The Nature of Adult Twin Relationships: An Attachment-Theoretical Perspective

Abstract

Twin relationships have been hailed as one of the most unique and intimate kinds of relationships. Unfortunately, there is a paucity of empirical research that addresses the interpersonal nature of twin relationships. In this article, the authors argue that attachment theory may provide a useful framework for understanding the nature of twin relationships. The authors present data indicating that (a) twins are more likely than non-twin siblings to use their sibling as an attachment figure; (b) the developmental course of twin attachment differs from that of other attachments; and (c) certain factors, such as genetic relatedness, empathy, including the other in the self, and shared experiences, may impact the extent to which twins use one another as attachment figures.

Full text

INTERPERSONAL RELATIONS AND GROUP PROCESSES
The Nature of Adult Twin Relationships:
An Attachment-Theoretical Perspective
Caroline M. Tancredy
University of California, Davis, and University of Illinois at
Urbana–Champaign
R. Chris Fraley
University of Illinois at Urbana–Champaign
Twin relationships have been hailed as one of the most unique and intimate kinds of relationships.
Unfortunately, there is a paucity of empirical research that addresses the interpersonal nature of twin
relationships. In this article, the authors argue that attachment theory may provide a useful framework for
understanding the nature of twin relationships. The authors present data indicating that (a) twins are more
likely than nontwin siblings to use their sibling as an attachment figure; (b) the developmental course of
twin attachment differs from that of other attachments; and (c) certain factors, such as genetic relatedness,
empathy, including the other in the self, and shared experiences, may impact the extent to which twins
use one another as attachment figures.
Twin relationships have captivated the public imagination for
centuries. From classic literature to modern media, twinships have
been portrayed as one of the most unique and intimate of inter-
personal bonds (Burlingham, 1952; Koch, 1966; Neyer, 2002b;
Segal, 1997, 1999; Woodward, 1998). Although there is no short-
age of psychological studies that employ twin samples, few of
these studies have addressed the nature of the twin relationship
itself. In this article, we argue that attachment theory (Bowlby,
1969; Hazan & Shaver, 1987) may offer a valuable framework for
understanding the nature of twin relationships in adulthood. We
begin with a brief review of attachment theory, focusing on the
features that differentiate attachment from nonattachment relation-
ships. On the basis of our review of existing twin research, we
suggest that twin relationships meet many of the criteria of attach-
ment relationships: They are characterized by proximity seeking,
separation distress, and the use of one another as a safe haven and
secure base. To empirically evaluate this hypothesis, we report a
study designed to compare the extent to which adult twins and
nontwins use their siblings as attachment figures.
1
We also explore
some of the factors that may influence the extent to which twin and
nontwin siblings use one another as attachment figures, such as
feelings of empathy, shared activities, the inclusion of the other in
the self, and genetic relatedness.
Although twins have figured prominently in the history of
psychological science, their role has primarily been to serve as a
tool for behavioral genetic research (e.g., Bouchard, 2004). It is
our hope that this article will help inspire a shift toward the
psychological study of the relationship between twins as a phe-
nomenon worthy of investigation in its own right (see Neyer,
2002b; Segal, 1999). Such research would prove valuable not only
for helping us better understand the nature of the twin bond, but
also for refining and extending contemporary theories of how close
relationships develop, function, and shape human experience.
Attachment Theory and the Attachment Behavioral
System
During the early stages of life, infants exhibit a strong propen-
sity to seek and maintain contact with a limited set of caregivers.
Bowlby (1969) hypothesized that such tendencies are the result of
a motivational system—the attachment behavioral system—that
1
Although the majority of research conducted on adult attachment is
concerned with individual differences in attachment organization, or “at-
tachment styles,” the primary objective of the research reported here is to
understand the development of attachment bonds—regardless of whether
those relationships are secure or insecure. In doing so, we are drawing on
a long-standing theoretical distinction between attachment as a normative
phenomenon (i.e., an affectional bond that characterizes the relationship
between two people) and attachment as an individual difference variable
(i.e., a style of relating in intimate relationships, see Fraley & Davis, 1997;
Hazan & Shaver, 1990; Hazan & Zeifman, 1994). We return to the question
of attachment styles in the General Discussion.
Caroline M. Tancredy, Department of Psychology, University of Cali-
fornia, Davis, and Department of Psychology, University of Illinois at
Urbana-Champaign; R. Chris Fraley, Department of Psychology, Univer-
sity of Illinois at Urbana-Champaign.
We thank Katherine Conger, Jennifer Frei, Ben Hankin, Kate Isaacson,
Joel Johnson, Glenn Roisman, Phil Shaver, Rick Robins, Kali
Trzesniewski, and Keith Widaman for constructive comments on drafts of
this article. We also thank the National Organization for Mothers of Twins
Clubs for their support.
Correspondence concerning this article should be addressed either to
Caroline M. Tancredy or R. Chris Fraley, Department of Psychology,
University of Illinois at Urbana–Champaign, Champaign, IL 61820.
E-mail: tancredy@uiuc.edu or rcfraley@uiuc.edu
Journal of Personality and Social Psychology, 2006, Vol. 90, No. 1, 78 –93
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.90.1.78
78

functions to regulate proximity between an infant and the primary
caregivers. According to Bowlby, the evolution of such a system
would have been critical to the survival of species that were born
with limited capacities for feeding, exploration, and defense.
Hence, infants are born with a set of innate behavioral responses
(such as crying and searching) and features (such as large eyes)
that attract the attention of potential caregivers. As children de-
velop, they begin to use goal-directed strategies for maintaining
proximity to an attachment figure. When the attachment figure is
perceived as distant or inaccessible, the child actively searches for
the parent or may resist separation by crying and clinging. In
contrast, when the attachment figure is nearby or accessible, the
infant experiences “felt security” (Sroufe & Waters, 1977), is more
sociable, and is more willing to explore the environment freely.
Although Bowlby was focused primarily on understanding the
nature of the infant– caregiver relationship, he believed that attach-
ment characterized human experience from the cradle to the grave.
It was not until the mid 1980s, however, that researchers began to
take seriously the possibility that the attachment system may be
operative in adulthood. Hazan and Shaver (1987) were two of the
first researchers to explore Bowlby’s ideas in the context of
adulthood, specifically, in adult romantic relationships. According
to Hazan and Shaver, the emotional bond that develops between
romantic partners is partly a function of the same motivational
system—the attachment behavioral system—that gives rise to the
emotional bond between infants and their caregivers. Since Hazan
and Shaver’s (1987) groundbreaking work, researchers have dis-
cussed the possibility that other kinds of adult relationships may
serve as attachment relationships, including friendships (Fraley &
Davis, 1997); relationships with family members, such as siblings
and grandparents (Ainsworth, 1989; Trinke & Bartholomew,
1997); and relationships with God (Kirkpatrick, 1995; 2004).
The Features of an Attachment Bond
Attachment relationships are characterized by four features or
functions (Ainsworth, 1991; Hazan & Zeifman, 1994). First, an
attachment figure is used as a target for proximity maintenance.
Both infants and adults enjoy being in the presence of their
attachment figures and actively seek those figures when they
accomplish something or when they feel threatened. Second, both
infants and adults experience separation distress when the rela-
tionship is disrupted. Separations from nonattachment figures gen-
erally do not elicit distress or vigorous attempts to reestablish
contact. Third, the attachment figure serves as a safe haven. When
a person is distressed, he or she often seeks the attachment figure
for contact, assurance, and safety. Although distressed infants may
accept care from other adults, infants are rarely soothed effectively
unless they are in the presence of a primary attachment figure
(Bowlby, 1969; Heinicke & Westheimer, 1965). Fourth, people
use their attachment figures as a secure base from which to explore
the world. Young children, for example, are comfortable with the
exploration of strange new environments only to the extent to
which they know that the attachment figure is nearby and acces-
sible if needed. Similarly, adult romantic partners are often much
more comfortable exploring careers and leisure activities when
they know that their partner is accessible.
Although attachment relationships are expected to be “close”
relationships, theoretically, there is no reason to assume that a
close relationship must be an attachment or that an attachment
relationship is necessarily a close or satisfying one (Cassidy,
1999). In other words, attachment relationships are assumed to be
a different kind of relationship, one that serves psychological
functions that typically are not served by other highly interdepen-
dent relationships (Ainsworth, 1991; Cassidy, 1999; Weiss, 1991).
A person, for example, may have a satisfying relationship with his
or her roommate, but may not use that person as a secure base from
which to explore the world. Alternatively, someone may be at-
tached to a parent but find the relationship to be an insecure
one— one in which the person’s needs are repeatedly unmet. In
theory, close relationships and attachment relationships are distinct
(Weiss, 1991), despite the fact that, in practice, attachment rela-
tionships are likely to be close and satisfying ones.
Attachment Hierarchies
In infancy as well as in adulthood, people are likely to have
multiple attachment figures. Bowlby (1969) discussed the notion
that infants could have multiple attachment objects and that “re-
sponsiveness to crying and readiness to interact socially are among
the most relevant variables” (p. 315) in determining who will be
chosen as an attachment figure. The mother is the most likely
candidate in most cases, but fathers, siblings, aunts, uncles, and
grandparents may also serve as attachment figures (Ainsworth,
1991).
The infant’s various attachment figures typically form what has
been called a “hierarchy” of attachment figures (Collins & Read,
1994; Trinke & Bartholomew, 1997). Infants do not treat all
attachment figures as equivalent; some are preferred over others.
Most infants prefer to seek comfort from their mothers when
distressed; in the mother’s absence, however, infants may look to
other attachment figures to meet their attachment needs. Bowlby
(1969) used the term monotropy to denote the tendency for infants
to prefer one attachment figure over others for comfort and
security.
As a child grows older, he or she has an opportunity to expand
his or her network of attachment relationships. Trinke and Bar-
tholomew (1997) extended the work of Hazan, Hutt, Sturgeon, and
Bricker (1991) by developing a self-report measure to assess
attachment hierarchies. Trinke and Bartholomew (1997) found that
most young adults have multiple attachment figures, including
family members, romantic partners, and friends. Mothers seem to
be given special status as attachment figures, followed by romantic
partners and best friends. Siblings were chosen less often than
partners, mothers, and fathers. These and other studies (e.g.,
Doherty & Feeney, 2004; Fraley & Davis, 1997) show that when
young adults become involved in romantic relationships, their
romantic partners typically move to the top of the existing attach-
ment hierarchy, with the exception of mothers, who may continue
to occupy a privileged position in the hierarchies of many young
adults.
Attachment in Twin Relationships
Part of what has been missing in the study of twins is a
theoretical model for understanding the nature of the bond between
them. In this section, we argue that attachment theory provides a
valuable framework for understanding the ways in which twins
79
ATTACHMENT IN TWIN RELATIONSHIPS

relate to one another. We begin by explaining the developmental
context of twinship and why that context may facilitate the estab-
lishment of attachments between twins. Next, we review briefly
existing qualitative and quantitative research on twin relationships,
emphasizing the findings as they bear on the key markers of an
attachment relationship: proximity seeking, separation distress,
and the use of a relationship partner as a safe haven and a secure
base. The primary purpose of this review is to demonstrate the
tenability of the idea that twinships may be attachment relation-
ships, but we also focus more generally on the notion that sibling
relationships may sometimes function as attachments.
Twin Development and Attachment
Like all infants, twins use a variety of behaviors, such as
locomotion and vocalization, to engage a particular figure who will
provide safety and security. These behaviors result in what Ains-
worth, Blehar, Waters, and Wall (1978) called “clear-cut attach-
ment” around the age of 6 –8 months At this stage of development,
an infant does not easily accept substitutes for the primary attach-
ment figure and is particularly sensitive to the figure’s accessibility
and responsiveness. In the case of twins, the mother must simul-
taneously meet the needs of two infants. As such, twins almost
always receive less individual attention from their mothers than
singleton children (Lytton, 1977). In fact, some research suggests
that mothers speak to their twin infants only half as often as they
speak to their singleton infants (Lytton, 1977; see also Ainslie,
1997). Therefore, as Ainslie notes, “at the developmental juncture
when the mother is uniquely important to her infant, the mother of
twins is most likely to feel overwhelmed by the demanding task of
meeting the needs of two infants at the same time” (Ainslie, 1997,
p. 22). Further, the added demands required by the “twin situation”
may alter the mother–infant relationship in important ways during
the first year of life (Ainslie, 1997).
Twin infants clearly seek proximity to their mother and attempt
to receive the same amount of interaction with her as do singleton
children. Also, during the first couple of years of life, they appear
to be relatively uninterested in the presence of the cotwin, although
the presence of the cotwin may have a soothing effect (Leonard,
1961). In other words, the presence of the cotwin does not dimin-
ish the need for the mother, nor does the twin prefer the company
of the cotwin to that of the mother. Savic (1980) observed 3
singletons and 3 sets of twins in their homes over the course of 8
months and found that toddler twins showed a clear preference for
their mother over their cotwin. Savic (1980) observed similar
behaviors among twins between the ages of 14 and 36 months;
they tended to interact with each other only if the mother or other
social partners were unavailable.
Twins begin to interact with each other in a relationship-
oriented manner around the age of 36 months, following the end of
the separation-individuation process (Mahler, Pine, & Bergman,
1975) at which time the child shifts from having an extremely
limited awareness of self and the outside world to experiencing
him- or herself as a separate, integrated person. Part of this process
involves an increasing interest in social partners and objects other
than the mother. Although the mother herself is still meticulously
scrutinized and explored, she becomes a secure base from which
the child can explore the environment and the child begins to
actively investigate other aspects of the world. In the case of twins,
the cotwin is a readily accessible object or partner. It is during this
important period of development that twins become increasingly
involved with one another (see Ainslie, 1997, for an in-depth
discussion of these developmental processes among twins).
Siblings as Attachment Figures and Factors That May
Promote Sibling Attachment
Although siblings are generally not used as attachment figures
in infancy or adulthood (Trinke & Bartholomew, 1997), there is no
theoretical reason why they cannot hold a special place in a
person’s attachment hierarchy (Ainsworth, 1991), especially
among twins. In fact, Ainsworth (1991) suggested that not only are
sibling attachments possible, but they can be facilitated by several
factors, including the inaccessibility of other attachment figures,
the trust that is developed through play interactions, and shared
experiences.
When the mother is perceived as inaccessible, siblings may turn
to each other for security, comfort, and care. Older siblings, who
may be more likely to play a parental, caregiving role with their
younger siblings, may be available to serve as supplementary
attachment figures for them, providing security and comfort when
they are needed (Ainsworth, 1991). Stewart (1983) studied 3- and
4-year-old children who were left alone together in a waiting room
without their mothers. He observed that half of his sample pro-
vided assurance and care to their younger siblings (see also Stew-
art & Marvin, 1984). These observations suggest that a child as
young as 3 or 4 years of age may serve as an attachment figure to
a younger sibling when the primary caregiver is unavailable.
In addition to the inaccessibility of other attachment figures,
Ainsworth (1991) suggested that siblings who are close in age may
form attachment relationships because they are likely to be play-
mates and to develop relationships characterized by cooperation,
reciprocity, and mutual trust. These qualities are common in at-
tachment relationships and may contribute to the formation of
secure attachment in sibling friendships (Fraley & Davis, 1997).
Further, shared background experiences may promote attachment.
Although siblings experience differences with respect to particular
activities and interests, their long history of shared experience
promotes similarities in their perception of situations and in value
systems that influence their decisions and mutual understanding.
This mutual understanding may be especially strong among twins,
who are the same age and are likely to have had a greater propor-
tion of shared experiences. Identical twins, in particular, are en-
couraged by parents and others to be similar and to stick together
(Koch, 1966; see Vandell, 1990).
Siblings may also become attached to each other for more distal
reasons. According to biological perspectives on kin relationships,
people are predisposed to behave in prosocial ways toward others
with whom they share genes (Hamilton, 1964; Neyer & Lang,
2003). On average, siblings share 50% of their genes with one
another. As such, directing attachment behavior toward a sibling
(as opposed to a nonfamily member) may benefit one’s inclusive
fitness. This is especially likely to be true for identical twins, who
share all of their genes. In fact, previous research has shown that
identical twins grieve the loss of their cotwin with more intensity
than do fraternal twins (Segal & Ream, 1998).
80
TANCREDY AND FRALEY

Features and Functions of Attachment in Twin
Relationships
Although there is evidence that twins often develop close,
interdependent relationships with one another, it is unclear whether
these relationships qualify as attachments per se. In the sections
below, we review both qualitative and quantitative research that
suggests that twins may serve as attachment figures for one an-
other across the life span. We will organize our discussion with
respect to the key features and functions that are thought to define
attachment relationships (Hazan & Zeifman, 1994): proximity
seeking, separation distress, the use of the other as a safe haven,
and the use of the other as a secure base.
Proximity seeking. In a large study of preschool-aged twins,
Koch (1966) found that twins were more likely than nontwins to
spend time together and share playmates. Whereas nontwin sib-
lings said they were happier without their brothers or sisters, twins
said they preferred being together. In fact, most of the twins
interviewed by Koch expressed a desire to be in the same class-
room. In a naturalistic observation study of twin pairs on a school
playground, Segal (1999) found that identical twins showed
greater physical closeness than fraternal twins. According to Segal:
When they were separated they could be seen looking around for one
another every once in awhile. The slight tensions in their faces when
they searched disappeared once they caught sight of their twin. This
type of behavior is like interactional melodies, the finely tuned rela-
tionship patterns, observed between parents and children in parks and
playgrounds around the world, in which separations are followed by
searches, and eventually, joyful reunions. (p. 105)
Less is known about proximity-seeking behaviors in adult twins,
but there is some evidence to suggest that twins are interested in
keeping the cotwin nearby and accessible. In a study of adult
twins, Tambs, Sundet, and Berg (1985) found that identical twins
were psychologically closer to each other and they lived together
longer than fraternal twins. In a study of twins in old age, Neyer
(2002b) found that twins contacted each other more frequently and
lived in closer geographical proximity than nontwin siblings.
Ainslie (1997) interviewed adult twins who described their
reactions to being separated from their cotwin and noted differen-
tial reactions among twins to various kinds of separations, such as
one twin taking a trip, having a steady boyfriend or girlfriend, or
marrying (see also Case, 1991). One interviewee, Cindy, was
particularly concerned about the accessibility of her identical twin
sister, Lindy. She stated “We were separated in college and I didn’t
know where she was, and she didn’t know where I was most of the
time. And it was at times when I really wanted to talk to her and
I just couldn’t find her anywhere that I would get really upset, like
‘Where is she!’. . . I would look and look, and sometimes I would
ask so-and-so, and then I would go there and couldn’t find her....
Somebody should know where she is! Somebody should try to help
me!” (Ainslie, 1997, pp. 75–76).
Separation distress. As implied by the previous quote, some
twins can be easily distressed when separated from one another.
This distress is particularly obvious when one twin dies (Wood-
ward, 1998). Segal and her colleagues (e.g., Segal & Bouchard,
1993; Segal, Wilson, Bouchard, & Gitlin, 1995) have conducted
extensive research showing that bereaved twins experience grief
more intensely for the loss of their cotwin than for other people
and that the intensity of their grief is greater than that experienced
by bereaved spouses. Some twins report that seeing the twin die
felt like experiencing their own death (Woodward, 1998). Some
twins express a sense of loneliness and loss so intense that nobody
can help to ease it (Woodward, 1998). Eighty-one percent of the
twins in Woodward’s (1998) sample indicated that the loss was
“severe” or “marked.”
Safe haven. Twins appear to place a high priority on offering
support, protection, and comfort to their cotwin in times of need
and regard the solace of the cotwin as irreplaceable. In fact, twins
are sometimes able to soothe each other even when others cannot
(Leonard, 1961). Tancredy (1999) reported on the case of Mike, a
23-year-old identical twin who commented about his cotwin, “He
always knew exactly how I felt. He was always the one person that
I could turn to if I had problems, if I was scared, if I was nervous.
I’d just turn to him and he knew. . . So if my dad would yell at us
or something, we’d always buddy up and go into a room and play
and we’d be fine.”
Lassers and Nordan (1978) suggest that the twin relationship is
an enduring attachment and that the twin naturally retreats to the
twinship for safety and security. The tendency to “fall back” on the
twinship as a source of safety and security was evident in the
interviews conducted by Tancredy (1999). For example, one fe-
male identical twin commented:
I think that when we were little, we used to be a lot closer. We used
to do everything together. Now that we’re older we don’t do every-
thing together. We have our own lives, so we have become more
independent. It’s more like the other one is a fall back. Like I can
always fall back onto my [twin] and she can always fall back onto me.
Before it was always ‘us.’. . . and now that we know other people, we
are kind of like a fall back.
Secure base. There is very little research to determine the
extent to which twins serve as secure bases for each other across
the life span. In early childhood, it seems likely that twins use their
cotwins as a secure base when they are psychologically separated
from the mother. Sandbank (1999) suggests that the twin, serving
as a “transitional object” in place of a teddy bear or comfort
blanket, may become an alternative secure base. However, this
base does not necessarily promote exploration. A base that has a
tendency to move with one may inhibit exploration by focusing
attention on the base rather than the territory to be explored.
Moreover, when neither twin has had the security of being able to
command his or her mother’s full attention, one twin may be less
willing to let the mother go unless accompanied by the other twin.
Other studies suggest that exploration need not be inhibited by
using one’s twin as a secure base. Koch (1966) found that children
who reported feeling close to their cotwin were less apprehensive
socially. Their parents reported that the twins were affectionate
and responsive. Others have also observed that preschool-aged
twins serve as attachment or security figures, defining attachment
in terms of apparent comfort in the presence of the attachment
figure and distress when separated. In a classic case study of a set
of Russian twins, Luria and Yudovitch (1959) noted that the
preschool-aged twins were happy and energetic when they were
together but were restrained and quiet when apart. In the same
vein, mothers have reported that their young twins worry about
each other when they are separated and help ease each other into
new situations (Ainslie, 1997).
81
ATTACHMENT IN TWIN RELATIONSHIPS

Overview of the Present Research
The studies reviewed previously suggest that twins may use
their cotwin as attachment figures. Nonetheless, there have been
no systematic attempts to evaluate this hypothesis among adult
twins. The objective of this research is to assess the presence of
attachment features and functions in twin relationships. For a point
of comparison, we focus on differences between the sibling rela-
tionships of twins and nontwins. The central questions motivating
this study are: (a) Are twins more likely than nontwin siblings to
use their siblings as attachment figures? (b) Where are siblings
positioned in the attachment hierarchies of twins and nontwins? (c)
What kinds of psychological and developmental characteristics
contribute to the development of an attachment bond with a
sibling? On the basis of the literature reviewed previously, we
hypothesized that twins would be more likely than nontwins to
regard their sibling as an attachment figure, and that, among twins,
cotwins would be placed at the top of the attachment hierarchy. We
also hypothesized that attachment bonds would be more salient
when (a) siblings had less supportive and trusting relationships
with their mothers, (b) the age difference between siblings was
small, (c) siblings shared many experiences while growing up, (d)
siblings experienced more empathy and closeness to one another,
and (e) siblings were more genetically related to one another.
Method
Participants
Sixty-two twins (30 identical, 32 fraternal) and 928 nontwins partici-
pated in this research. Only one member of a twinship or sibling relation-
ship was needed to complete the study. Participants were drawn from both
a student population (in exchange for course credit) and the Internet
community (in exchange for educational feedback about the participant’s
close relationships). Participants ranged in age from 14 to 61 years (M ⫽
23, SD ⫽ 7.8). The majority were students who participated for course
credit (60%). About 78% of participants reported annual incomes under
$20,000, 10% reported annual incomes between $20,001 and $40,000, 4%
reported incomes between $40,001 and $60,000, and the remaining 6%
reported incomes above $60,001. Regarding ethnicity, 60% of the sample
reported that they were White. Additionally, 3% were African American,
19% were Asian/Asian American, 5% were Pacific Islander, 7% were
Hispanic, and 3% were Middle Eastern. The remaining 3% reported
“other” or did not answer. Eighty percent of the participants were women.
In this study, participants were required to answer questions about a
sibling. For the nontwin sibling participants, 53% of the sample described
a sibling who was younger, and 46% described a sibling who was older.
The absolute age difference between siblings ranged from 1 to 22 years
(M ⫽ 4.1). Fifty-three percent of the nontwin siblings were women. In the
twin group, 68% of the cotwins were women. Twenty-three percent of the
sample reported that they currently lived in the same household as their
sibling.
Procedure
Participants completed a set of questionnaires on the Internet as part of
an online study of sibling relationships. The study was posted on a Web site
hosted by R. Chris Fraley called yourPersonality.net. This site contains a
variety of online tests for issues of relevance to personality, attachment,
close relationships, and politics. The majority of people who voluntarily
visit the site are women (80%), probably because women tend to be more
interested in relational issues than men (e.g., Martin, 1991). It is important
to note, however, that women were neither more nor less likely to partic-
ipate in the current research than that for other projects posted on that Web
site. Participants were asked to register on the study Web site using a
username and password. They had the option of completing the question-
naire all at once or returning to the site later and completing the question-
naire at their convenience. Once all the measures were completed, partic-
ipants were thanked, debriefed, and given extensive personal feedback that
was automatically generated from the participants’ responses.
Measures
Family questions. Participants answered several questions about their
family. Participants were asked to report their family size while they were
growing up. They also indicated their relation to the individuals they
regarded as their mother and father (biological, adopted, step, foster, or
other).
Sibling questions. Participants were asked to report the names, birth
dates, and ages of their siblings and the sibling relation for each one (full,
half, step, or adopted). This information was used to determine age differ-
entials for nontwin siblings and the proportion of genes shared between
siblings. We also asked nontwin respondents to “choose a sibling that you
would like to consider for the rest of this study” and to indicate the first
name of that sibling. The remaining questions about siblings (see below)
were asked in reference to that particular sibling. We did not provide
special directives to respondents on how to choose which sibling, if he or
she had more than one, to nominate.
2
Zygosity. Participants who indicated that their sibling was a twin were
also asked to self-report their twin type (identical, fraternal, or don’t know)
and to explain how they came to know their type (e.g., blood test, cheek
swab test, “doctor told me,” or “parents told me”). We used siblings’
self-reported twin type to classify the zygosity of participants. Siblings who
were unsure of their twin type were excluded from the analyses.
3
Attachment features and functions. To assess attachment-related func-
tions and features, we administered a modified version of the WHOTO
Questionnaire (Fraley & Davis, 1997; Hazan et al., 1991) and the Attach-
ment Network Questionnaire (ANQ; Trinke & Bartholomew, 1997). The
measure included 16 items that assessed the degree to which a potential
attachment figure (sibling, mother, father, romantic partner, or friend)
served each of the four primary attachment functions. We constructed these
scales based on attachment theory and research, choosing items that
seemed to best represent these functions (see Appendix). Most items were
taken from established attachment measures (e.g., the WHOTO and the
ANQ). For example, proximity maintenance items included such state-
ments as “I make an effort to stay in contact with my sibling.” The
separation distress items included such statements as “When I am away
from my sibling, I feel down.” The safe haven subscale included items such
as “My sibling is a person I count on for advice.” Finally, the secure base
subscale was composed of items such as “My sibling is the person that I
2
We did not include directives because we wanted to make sure that
nontwin siblings had the opportunity to rate a sibling who may be impor-
tant to them. From this perspective, our design is a rather conservative one
because it would undermine our ability to detect a difference between twin
and nontwin siblings in the extent to which they use their siblings as
attachment figures.
3
Zygosity was also assessed with a four-item measure, developed by
researchers at the University of Minnesota, that has been shown to assess
zygosity accurately in 96% of cases (Peeters, Van Gestel, Vlietinck,
Derom, & Derom, 1998). These items require participants to indicate
similarity and difference characteristics that help to determine genetic
similarity. For example, one item reads, “During your childhood, were you
and your twin as alike as ‘two peas in a pod,’ or were you no more alike
in appearance than ordinary brothers and sisters?” The point-biserial cor-
relation between the self-report classification of zygosity and a unit-
weighted composite of the Peeters et al. items was .78.
82
TANCREDY AND FRALEY

count on to always be there for me and care about me no matter what.”
Participants were asked to rate the degree (from 1, strongly disagree,to7,
strongly agree) to which attachment features and functions characterize
their relationships with particular people (i.e., parents, partners, and twin or
nontwin siblings) rather than nominate specific individuals or provide
binary responses (see Fraley & Davis, 1997, for an example of a nomina-
tion procedure). We should note that, although attachment theorists often
speak about an attachment relationship as either existing or not (e.g.,
Cassidy, 1999), we assessed the extent to which the relationship resembled
an attachment relationship quantitatively. We did so because continuous
information can always be used to make classifications, if theory demands,
but categorical data cannot easily be used to make graded distinctions.
Although we rely on continuous ratings in the current report, we remain
agnostic on the thorny issue of whether attachment relationships are truly
“all or none” affairs.
Inclusion of other in self. The Inclusion of Other in Self measure is a
single item designed to tap people’s sense of being closely interconnected
with another person (IOS; Aron, Aron, & Smollan, 1992; Aron, Aron,
Tudor, & Nelson, 1991). The IOS was designed as a measure of closeness,
but it is also meant to reflect the fact that closeness involves a complex
sharing of characteristics, resources, and perspectives. In a close relation-
ship, an individual acts as if some or all aspects of the relationship partner
are also the individual’s own (Aron et al., 1992). A higher IOS score
indicates a greater sense of interconnectedness with the target person.
Participants in the present study were asked to choose one of several sets
of Venn diagrams that best reflected the degree of overlap between self and
other with respect to their relationship with their sibling, mother, father,
friend, and romantic partner.
Quality of relationship with mother. To assess the quality of the
relationship with mother, participants were asked to rate ona1(strongly
disagree)to7(strongly agree) scale the extent to which each of the
following adjectives characterize their relationship with their mother: hon-
est, caring, trusting, supportive, respectful, understanding, affectionate,
accepting. These ratings were averaged to create a composite index of the
quality of the relationship with mother (
␣
⫽ .91).
Quality of relationship with sibling. To assess the quality of the
relationship with their sibling, participants were asked to rate on a scale of
1(strongly disagree)to7(strongly agree) the extent to which each of the
following adjectives characterize their relationship with their brother or
sister: honest, caring, trusting, supportive, respectful, understanding, affec-
tionate, accepting. These ratings were averaged to create a composite index
of the quality of the relationship with the sibling (
␣
⫽ 92).
Twin and Sibling Relationship Inventory. A Twin and Sibling Rela-
tionship Inventory (TSRI) was constructed specifically for this study to
gain information about the nature of the relationship between the partici-
pant and his or her cotwin or sibling. Items on the TSRI were constructed
from qualitative reports of relationships with twins (e.g., Ainslie, 1997;
Schave & Ciriello, 1983) and siblings (e.g., Bank & Kahn, 1982). The
TSRI contains items designed to tap the kinds of themes that have emerged
from the twin literature. We used these items to create a measure of the
degree of empathy experienced in the sibling relationship (“When some-
thing bad happens to my sibling, I feel bad too”; 6 items;
␣
⫽ .83); the
extent to which siblings shared common experiences and activities while
growing up (“Growing up, my sibling and I were encouraged to share
everything”; 4 items;
␣
⫽ .74); and the degree to which siblings currently
have separate lives (“My sibling and I have separate personal lives”; 4
items;
␣
⫽ .73).
Results
Attachment Functions Measure
Because the measure of attachment functions used in this study
was novel, our first step was to explore its psychometric proper-
ties. The correlations among the four scales, as well as the means,
standard deviations, and alphas, are reported in Table 1. The
internal consistencies for these scales within a target relationship
were high, ranging from .84 for the separation distress scale to .90
for the safe haven scale (see Table 1). Within each relationship
type (i.e., sibling, mother, father, romantic partner, and friend), the
four scales were highly intercorrelated (on average, .75). The high
magnitude of these associations indicates that there was a single
source of variation underlying the attachment function ratings.
Indeed, the results of principal axis factor analyses (using oblique
rotation) within each relationship domain indicated that there was
one principal factor underlying the covariation among items (e.g.,
in the sibling domain, the first four eigenvalues were 10.95, 1.85,
1.08, and .90). Conceptually, this factor represents the extent to
which a target is regarded as an attachment figure. Because this
factor captured an overwhelming majority of the variance, in our
primary analyses we focus on a unit-weighted composite index
that captures this major source of variation within each relation-
ship domain. This composite captures the extent to which the
subject relies on the target as an attachment figure (i.e., uses the
person as a target of proximity maintenance, safe haven, and
secure base, and experiences separation distress in the person’s
absence). The means, standard deviations, and alphas for the
composite attachment scales across different kinds of relationships
are presented in Table 2.
4
Are Twins More Likely Than Nontwins to Rely on Their
Sibling as an Attachment Figure?
To determine whether twins are more likely than nontwins to
use their siblings as attachment figures, we compared the average
sibling attachment composite score for twins and nontwins. The
means and standard deviations for attachment scores (for twins and
nontwins) across relationships are presented in Table 3. Overall,
twins (M ⫽ 5.35, SD ⫽ 1.22) were more likely than nontwins
(M ⫽ 4.53, SD ⫽ 1.34) to regard their sibling as an attachment
figure, t(981) ⫽ 4.52, p ⬍ .05, d ⫽ .64.
Do Twins and Nontwins Differ in Their Regard for
Others as Attachment Figures?
We also sought to determine whether twins and nontwins dif-
fered in their use of nonsiblings (e.g., mothers, friends) as attach-
ment figures. The means and standard deviations for each rela-
tionship are reported in Table 3. Twins were less likely than
nontwins to regard mothers, t(986) ⫽⫺3.67, p ⬍ .05, d ⫽⫺.46,
as attachment figures. A similar trend was observed with respect to
fathers, t(978) ⫽⫺1.81, p ⫽ .07, d ⫽⫺.24, and friends, t(986) ⫽
⫺1.77, p ⫽ .07, d ⫽⫺.23. Finally, although twins did not report
being attached to their romantic partners to a lesser extent than did
nontwins in the full sample, t(978) ⫽⫺1.52, p ⫽ .12, d ⫽⫺.19,
when we conducted this analysis only on participants who were
involved in dating and marital relationships, the effect emerged,
4
Sex did not moderate any of the analyses we report. Moreover, twins
were no more likely to be male or female than were nontwins. The twins
in our sample, however, tended to be 3 years older, on average, than
nontwins. When we reran our key analyses (e.g., those summarized in
Table 2) controlling for age, our results were unchanged.
83
ATTACHMENT IN TWIN RELATIONSHIPS

Table 1
Correlations, Means, Standard Deviations, and Alphas Between Attachment Functions Scales for All Relationships
1234567891011121314151617181920MSDa
Attachment to Sibling
1. Proximity Maintenance — 5.2 1.61 .85
2. Separation Distress .76 — 5.1 1.24 .84
3. Safe Haven .68 .68 — 3.75 1.7 .94
4. Secure Base .73 .73 .85 — 4.65 1.57 .85
Attachment to Mother
5. Proximity Maintenance .30 .28 .15 .21 — 5.73 1.46 .88
6. Separation Distress .24 .38 .19 .24 .77 — 5.6 1.25 .86
7. Safe Haven .20 .23 .20 .20 .74 .74 — 4.8 1.66 .93
8. Secure Base .20 .25 .15 .23 .79 .80 .85 — 5.62 1.41 .85
Attachment to Father
9. Proximity Maintenance .27 .19 .13 .18 .36 .27 .27 .30 — 4.9 1.81 .91
10. Separation Distress .19 .25 .15 .19 .29 .38 .28 .31 .81 — 5.2 1.49 .90
11. Safe Haven .16 .13 .21 .18 .22 .21 .35 .26 .70 .71 — 3.6 1.76 .94
12. Secure Base .15 .13 .11 .18 .24 .22 .26 .31 .80 .81 .81 — 4.7 1.74 .88
Attachment to Friend
13. Proximity Maintenance .18 .18 .04 .11 .25 .19 .13 .20 .22 .17 .06 .12 — 5.78 1.12 .78
14. Separation Distress .16 .27 .08 .14 .23 .31 .17 .23 .19 .26 .10 .13 .71 — 5.47 1.09 .85
15. Safe Haven .06 .14 .04 .09 .12 .16 .08 .15 .11 .14 .06 .08 .71 .74 — 5.33 1.31 .92
16. Secure Base .09 .14 .01 .10 .16 .17 .11 .19 .13 .14 .05 .10 .73 .76 .86 — 5.57 1.14 .84
Attachment to Partner
17. Proximity Maintenance .11 .08 .00 .05 .12 .10 .04 .10 .10 .07 ⫺.01 .03 .20 .15 .09 .14 — 6.27 1.03 .85
18. Separation Distress .07 .10 .02 .05 .09 .14 .05 .09 .09 .11 .03 .03 .14 .23 .10 .14 .76 — 5.96 1.05 .87
19. Safe Haven .05 .05 .00 .03 .02 .05 .02 .06 .02 .04 .01 .00 .07 .10 .04 .08 .73 .76 — 5.82 1.19 .92
20. Secure Base .07 .08 .00 .04 .04 .07 .03 .08 .06 .07 .01 .03 .14 .15 .10 .16 .75 .74 .97 — 5.92 1.08 .83
84
TANCREDY AND FRALEY

t(554) ⫽⫺1.98, p ⬍ .05, d ⫽⫺.33. Twins (M ⫽ 5.91, SD ⫽ .98)
were less likely to regard their romantic partners as attachment
figures than nontwins (M ⫽ 6.21, SD ⫽ .85). In summary, twins
were more likely than nontwins to regard their siblings as attach-
ment figures, and they tended to regard other people as attachment
figures to a lesser extent than did nontwin siblings.
How Are Twins and Nontwin Siblings Positioned in the
Attachment Hierarchy?
Although twins are more likely to use their sibling as an attach-
ment figure than are nontwins, this finding does not necessarily
imply that siblings are at the top of the attachment hierarchy for twins.
To examine the placement of siblings within the attachment hierarchy,
we conducted a repeated measures analysis to compare the attachment
means across the five attachment targets. Sibling status (i.e., twin vs.
nontwin) was treated as a between-subjects factor and target (i.e.,
sibling, mother, father, partner, or friend) was treated as a within-
subjects factor. The analysis yielded a main effect for target, F(4,
964) ⫽ 37.8, p ⬍ .05, and an interaction, F(4, 964) ⫽ 13.97, p ⬍ .05,
indicating that the pattern of means for twins differed from the pattern
of means for nontwins. The means and standard deviations for attach-
ment targets are presented in Table 3. As can be seen, partners,
siblings, and friends were rated the highest among twins. This finding
suggests that a twin’s cotwin does not exclusively hold the highest
position in the attachment hierarchy; statistically, the cotwin scores
just about as high as other important attachment figures. In contrast,
however, the pattern of means for nontwins shows that siblings are
placed at the bottom of the attachment hierarchy. With the exception
of the higher placement of siblings for twins, a similar ordinal pattern
emerged for nontwin participants for partners, friends, mothers, and
fathers, respectively.
The Development of Attachment for Twin and Nontwin
Siblings: A Cross-sectional Analysis
Thus far, the data suggest that twin and nontwin siblings differ in
the extent to which they use their siblings as attachment figures.
Specifically, twins are more likely to regard their sibling as an
attachment figure than are nontwins. Moreover, nontwins are more
likely to rely upon their parents for various attachment-related func-
tions than are twins. In the set of analyses to be reported next, we
sought to delineate further the differences in the ways in which twins
and nontwins approach different kinds of attachment relationships.
According to attachment theory and research, as people grow older
they are more likely to use their romantic partners for attachment-
related functions and less likely to rely upon their parents for these
purposes (Fraley & Davis, 1997). It is unclear, however, whether
sibling attachments are maintained or relinquished over time.
To investigate these issues, we took advantage of the cross-
sectional nature of the data to examine the association between age
and attachment to various targets (i.e., siblings, parents, and ro-
mantic partners) as a function of sibling status (i.e., twin or
nontwin). In each analysis, we hierarchically regressed the extent
to which the target was used as an attachment figure on sibling
status, age, and the interaction between sibling status and age. The
regression analyses are summarized in Figure 1. Regarding attach-
ment to siblings, there was a main effect of age and sibling status.
Overall, older participants reported feeling less attached to their
siblings (
␤
⫽⫺.17, p ⬍ .05), and twins reported feeling more
attached to their siblings than nontwins did (
␤
⫽ .15, p ⬍ .05).
These effects were qualified by an interaction between age and
sibling status (
␤
⫽ .12, p ⬍ .05). As can be seen in the upper-left
panel of Figure 1, older twins felt more attached to their cotwins
than did younger twins, whereas older nontwins felt less attached
to their siblings than did younger nontwins.
5
The association between attachment and age was quite different for
other kinds of relationships. As can be seen in the upper-right panel of
Figure 1, older participants reported feeling less attached to their
mothers (
␤
⫽⫺.30, p ⬍ .05), and twins reported feeling less attached
to their mothers than nontwins (
␤
⫽⫺.09, p ⬍ .05). There was no
interaction between age and sibling status (
␤
⫽ .01, ns). A similar
pattern of results was found with attachment to fathers (see the
lower-left panel of Figure 1). There was a main effect of age (
␤
⫽
5
To ensure that the cross-sectional analyses were not unduly influenced
by a select few older participants, we reran these and other cross-sectional
analyses with a truncated age range (i.e., from 18 to 40). Both the stan-
dardized and unstandardized regression coefficients did not change in any
appreciable manner. For example, the standardized coefficients for this
analysis for age, sibling status, and their interaction were ⫺.12, .13, and
.11, respectively, for the truncated sample.
Table 2
Means, Standard Deviations, and Alphas for the Extent to
Which Various Targets Were Used as Attachment Figures
MSD
␣
Composite Attachment Scale
Sibling 4.57 1.37 .95
Mother 5.39 1.33 .95
Father 4.59 1.54 .96
Friend 5.93 1.01 .95
Romantic Partner 5.48 1.06 .95
Note. For attachment to romantic partners, the mean, SD, and alpha for
the subsample of people in a dating or marital relationship was 6.19, .86,
and .96.
Table 3
Means and Standard Deviations for the Extent to Which Various
Targets Were Used as Attachment Figures for Twins and
Nontwins
Twin
(n ⫽ 62)
Nontwin
(n ⫽ 926)
Attachment target
Sibling 5.35
cd
(1.22)
4.53
a
(1.34)
Mother 4.80
a
(1.48)
5.44
b
(1.32)
Father 4.24
b
(1.69)
4.62
a
(1.53)
Friend 5.26
c
(1.03)
5.50
b
(1.07)
Partner 5.75
d
(1.07)
5.95
c
(.99)
Note. Means with different subscripts within columns represent signifi-
cant differences, p ⬍ .05.
85
ATTACHMENT IN TWIN RELATIONSHIPS

⫺.16, p ⬍ .05), but no main effect of sibling type (
␤
⫽ .05, ns)or
interaction between sibling status and age (
␤
⫽ .02, ns).
Finally, we examined the extent to which twin and nontwin
siblings regard their romantic partner as an attachment figure over
the course of their relationship. Because not everyone in our
sample was involved in a dating relationship, we limited these
analyses to the subsample of participants who were in exclusive
relationships and studied (the log of) relationship length rather
than age per se. There was a main effect of sibling status (
␤
⫽
⫺.10, p ⬍ .05) such that twins were less likely to feel attached to
their partners than nontwins. Also, older respondents tended to feel
more attached to their partners than younger respondents (
␤
⫽ .11,
p ⬍ .05). There was no interaction between sibling status and
relationship length (
␤
⫽⫺.03, ns). The lower-right panel of
Figure 1 illustrates these regressions.
Factors That Facilitate and Inhibit the Development of
Attachment
We next examined the associations between the extent to which
the sibling was used as an attachment figure and a variety of
psychological, social, and developmental factors. Because our
measure of sibling attachment was correlated highly with the
quality of the sibling relationship (r ⫽ .73, p ⬍ .05), we thought
it would be prudent to study the association between sibling
attachment and these variables while statistically controlling sib-
ling relationship quality. Such an analysis should enable us to
identify the association between attachment and the various rela-
tionship variables independently of the quality of the relationship.
The partial correlations (computed separately for the full sample,
twins, and nontwin siblings) are summarized in Table 4. As can be
Figure 1. Attachment to different targets (i.e., sibling, mother, father, and romantic partner) as a function of
age and sibling status. Solid lines correspond to twin siblings. Dashed lines correspond to nontwin siblings.
86
TANCREDY AND FRALEY

seen, people were more likely to regard their sibling as an attach-
ment figure if they were encouraged to spend time together as
children; they do not have separate interests, personal lives, and
professional lives as adults; they experience empathy for one
another; and they include the other as part of the self. It is
noteworthy that these patterns of associations held for both twins
and nontwins and that, with a few exceptions, they were consistent
with what might be expected given the existing literature on twins.
The exceptions are, first, that there was not a significant negative
correlation between the relationship siblings had with their mother
and the extent to which they were attached to their sibling. Second,
the age differential between siblings, as well as family size, was
unrelated to the extent to which the sibling was used as an
attachment figure. Apparently, psychological factors, such as em-
pathy and shared interests, matter more for the development of a
sibling attachment bond than structural factors, such as family size
and age differences between siblings.
Integrating Genetic and Relational Perspectives on Twin
Attachment
Up to this point our analyses have focused on the way in which
twins and nontwins differ from one another in their sibling rela-
tionships as if there was something about twinship per se—as a
relational phenomenon—that matters. It is worth noting, however,
that twins differ from nontwins not only in the nature of their
relationships (as we have documented) but also in their genetic
relatedness: identical twins share more genes than nontwins. More-
over, even twins differ from one another with respect to genetic
relatedness: identical twins share 100% of their genes whereas
fraternal twins share, on average, 50% of their genes. It is possible
that genetic relatedness is a critical factor in shaping the bond that
develops between siblings.
Although social psychological and evolutionary perspectives are
often placed in opposition to one another in contemporary psy-
chology, we believe that it is possible to integrate these seemingly
divergent perspectives on human nature. As many ethologists have
noted (e.g., Hinde, 1982; Tinbergen, 1963), evolutionary theories
often focus on ultimate mechanisms (e.g., factors that indirectly
influence inclusive fitness or the selection of specific genotypes),
whereas psychological theories often focus on proximate mecha-
nisms (e.g., factors that more directly influence behavioral phe-
nomena). The fact that these different kinds of explanations are
often poised at different levels of analysis suggests that they are
not mutually exclusive and that a comprehensive account of be-
havioral phenomena may require attention to each. In the follow-
ing analyses, we investigate the role that genetic similarity be-
tween siblings may play in the nature of the relationship that
develops between siblings. Although we will explore the role of
genetic relatedness, we emphasize the way in which genetic per-
spectives may be integrated with relational ones to provide a more
complete understanding of attachment dynamics.
For our first analysis we created a new variable— one that did
not perfectly confound genetic relatedness with sibling status (i.e.,
twin or not twin). Specifically, we created a quasi-continuous scale
using the following values: 0% for foster and adoptive siblings,
25% for step siblings, 50% for dizygotic twins and nontwin sib-
lings, and 100% for monozygotic twins. This measure of genetic
relatedness was positively correlated with the extent to which
people use their siblings as attachment figures (r ⫽ .15, p ⬍ .05).
The correlations between this measure and the various relational
variables are reported in the right-most column of Table 4. It is
noteworthy that genetic relatedness correlates with many of these
relational variables. Siblings who have more of their genes in
common are more likely to have had shared experiences while
growing up, share common interests now, empathize with one
another, and include the other into the self. They were also, to our
surprise, more likely to have positive relationships with their
mothers. Taken together, these findings are compatible with the
notion that genetic similarity may play a broad role in shaping the
nature of the relationship that develops between siblings.
6,7
Another way to examine these issues is by honing in on the
distinction between nontwin siblings and fraternal twins. From a
genetic perspective, these two kinds of relationships are identi-
cal—in both cases, siblings share an average of 50% of their genes.
From a relational perspective, however, fraternal twins and non-
twins are characterized by different kinds of experiences. Thus, if
6
It should be noted that the correlation between genetic relatedness and
age differential reflects nothing more than the fact that twin siblings (who
share more of their genes than nontwin siblings, on average) do not vary in
their age differentials, whereas nontwin siblings have variable age
differentials.
7
We should note that the question of how genetic relatedness impacts
attachment is distinct from the kind of question that is addressed in a
common behavior genetic study using a twin design (see Neyer & Lang,
2003). A typical behavior genetic study compares the similarity of
monozygotic (MZ) and dizygotic (DZ) twins to determine whether MZ
twins are more similar than DZ twins with respect to a trait. In this kind of
research, it is not relevant whether MZ twins have higher or lower trait
scores, on average, than DZ twins. The focus is typically on covariance
structures rather than on mean structures. In our research, we are trying to
determine whether MZ twins are more likely to be attached to one another
than are DZ twins (and, of course, how nontwin siblings fit into this
picture). The answer to this question is independent of the heritability
coefficient that is estimated in a typical behavior genetic study.
Table 4
Partial Correlations Between Relational, Developmental, and
Structural Variables and Attachment to Sibling for the Full
Sample, Twins, and Nontwins, Controlling for Sibling
Relationship Quality
Variables
Attachment to sibling
Genetic
relatedness
Full
sample Twins Nontwins
Positive relationship with
mother
⫺.02 ⫺.19 .05 .15*
Shared experiences growing
up
.25* .46* .25* .20*
Separate lives ⫺.23* ⫺.44* ⫺.17* ⫺.13*
Empathy .46* .49* .48* .14*
Inclusion of other in self .50* .55* .48* .15*
Age differential ⫺.01 NA ⫺.04 ⫺.31*
Family size ⫺.01 ⫺.05 .05 .00
Note. The correlation between age differential and attachment to sibling
is not reported among twins because there was no variance in the age
differential among twins. *p ⬍ .05
87
ATTACHMENT IN TWIN RELATIONSHIPS

fraternal siblings are more likely to use one another as attachment
figures than nontwin siblings, such a finding would suggest that
the twin experience might be relevant for the establishment of an
attachment bond above and beyond the role of genetic similarity.
Moreover, because both fraternal and identical twins are twins but
differ in their genetic relatedness, if identical twins were more
likely than fraternal twins to be attached to one another, such a
finding would implicate a unique role for genetic similarity in
shaping attachment dynamics. These kinds of predictions can be
formalized as follows: From a genetic perspective, we would
expect the means for attachment to sibling to be ordered in the
following manner for nontwin siblings (NTS), fraternal twins
(DZ), and identical twins (MZ): (NTS ⫽ DZ) ⬍ MZ. From a
relational perspective, the means should be ordered as NTS ⬍
(DZ ⫽ MZ). An integrated perspective would postulate roles for
both kinds of factors, leading to the following prediction: NTS ⬍
DZ ⬍ MZ.
We tested these three sets of predictions using a series of
planned comparisons with one-tailed tests. In the genetic compar-
ison, we tested the difference between the means of the NTS and
DZ groups and the MZ group. This contrast was significant,
t(980) ⫽ 3.73, p ⬍ .05, indicating that nontwin siblings were less
likely to be attached to their siblings than twins. The relational
contrast for NTS versus DZ and MZ twins was also significant,
t(980) ⫽ 4.57, p ⬍ .05, indicating that people who share 50% of
their genes, on average, with their siblings are less attached to their
siblings than people who share 100% of their genes with their
siblings. Finally, we tested the ordered pairwise differences among
groups. These analyses indicated that nontwins were less attached
to their siblings (M ⫽ 4.53, SD ⫽ 1.38) than fraternal twins (M ⫽
4.94, SD ⫽ 1.27; t(951) ⫽⫺1.65, p ⬍ .05, d ⫽⫺.31), who, in
turn, were less attached to their siblings than were identical twins
(M ⫽ 5.77, SD ⫽ 1.02; t(951) ⫽⫺2.82, p ⬍ .05, d ⫽⫺.72).
These data are best explained by a perspective that assumes a role
for both genetic and relational processes in the development of an
attachment relationship.
In the next set of analyses, we sought to examine the pathways
through which genetic relatedness might influence the nature of
the sibling bond. Specifically, we evaluated an integrated model
that assumes that genetic relatedness between siblings facilitates
the amount of time that siblings spend together, the number of
shared activities in which they engage, and the degree of empathy
and closeness that is experienced. These social–relational factors,
in turn, are theorized to shape the formation of the attachment bond
between siblings. To test this mediational model, we followed the
general guidelines outlined by Baron and Kenny (1986). Specifi-
cally, we first established that there were associations between
each of the mediators and genetic relatedness and between the
mediators and siblings attachment (see Table 4). Next, we re-
gressed sibling attachment on both the relational mediators and
genetic relatedness.
8
Each hypothesized mediator, considered sep
-
arately, played a significant role in mediating the relationship
between genetic relatedness and the extent to which the sibling
was used as an attachment figure. The Sobel z values for shared
experiences, separate lives, empathy, and inclusion of the other in
the self were 5.29, 3.13, 4.06, and 4.46, respectively (all ps ⬍ .05).
The results for the full multivariate mediational model are sum-
marized in Figure 2. It is important that the estimated effect of
genetic relatedness on sibling attachment was virtually zero (
␤
⫽
.04, ns) once accounting for the various relational factors. More-
over, allowing this parameter to be freely estimated as opposed to
constraining its value to zero did not lead to significant improve-
ment in the capacity of the model to account for the observed
correlations,
␹
2
(1) ⫽ 1.46, ns. In short, these findings are consis
-
tent with the notion that genetic similarity promotes certain rela-
tional dynamics (such as an increased sense of empathy and
self– other overlap) among siblings and that these factors, in turn,
may help to shape the bond that develops between them.
9
General Discussion
Twin relationships have been hailed as one of the most unusual,
intimate, and mysterious of interpersonal bonds (e.g., Burlingham,
1952; Koch, 1966; Segal, 1997; Woodward, 1998). Despite the
unusual features of twin relationships, few researchers have at-
tempted to study this kind of relationship systematically. The
objective of the present research was to conceptualize twin rela-
tionships within the broad framework provided by attachment
theory and to empirically test several hypotheses about twin rela-
tionships derived from this perspective.
According to attachment theory, an attachment relationship is
one in which a person uses the other as a target of proximity
maintenance, a safe haven during times of distress, and a secure
base from which to explore the world. To examine the extent to
which twins and nontwins use their siblings as attachment figures,
we asked people to rate the extent to which their siblings fulfill
attachment-related functions. Our data indicate that twins are more
likely than nontwin siblings to regard their sibling as an attachment
figure. Twins were also less likely than nontwins to use their
parents as attachment figures. Although twins were more likely
than nontwins to be attached to their sibling, they were not nec-
essarily more likely to be attached to their sibling compared with
other important people in their life (i.e., romantic partners and
friends). In other words, although cotwins were at the top of the
attachment hierarchy among twins, they shared that spot with other
significant people, such as friends and romantic partners.
Our cross-sectional analyses also revealed that twins and non-
twins differed in attachment as a function of age. Whereas older
twins were more likely to rely upon their siblings for attachment-
related functions than younger twins, older nontwins were less
likely to do so than younger twins. We also found that a number
of factors predict the extent to which a sibling is used as an
attachment figure. For example, people were more likely to use
their siblings as attachment figures if they had many shared expe-
8
We statistically removed sibling relationship quality from the sibling
attachment scores using the residuals from a regression analysis to hone in
more specifically on attachment as opposed to relationship quality. The
results are the same regardless of whether relationship quality is controlled.
9
Although we focus on “separate lives” as an index of psychological
distance between siblings, we also measured physical distance as the
number of miles between the respondent and his or her sibling. This
measure of distance was negatively correlated with sibling attachment (r ⫽
⫺.08, p ⬍ .05), suggesting that the further away two siblings live from one
another, the less likely they are to use one another as attachment figures.
The correlation between physical distance and genetic relatedness, how-
ever, was virtually zero (r ⫽ .02, ns), indicating that physical distance did
not mediate the association between genetic relatedness and attachment.
88
TANCREDY AND FRALEY

riences with them, did not currently lead separate lives character-
ized by separate interests and professions, experienced a high
degree of empathy for one another, and psychologically included
the other as part of the self. Contrary to predictions, factors such as
the age difference between siblings did not predict attachment.
Siblings who were close in age were just as likely as those who
were not to be attached to one another. Also, although previous
authors have suggested that a negative relationship with the mother
might promote sibling attachment, we did not find strong evidence
for this assumption.
Implications for Understanding Twin Relationships
This research was inspired by two observations. First, the kinds
of qualities that are claimed to be “special” about twin relation-
ships also appear to characterize other kinds of intimate relation-
ships. For example, there are many examples of older married
couples that have astonishingly similar interests, spend all of their
time together, organize their lives around each other, and are
dependable sources of comfort, nurturance, and support for each
other. If these relationships involve the same psychological dy-
namics as twin relationships, it would be worthwhile to adopt a
theoretical framework that could accommodate them both. Second,
as rich as the qualitative literature on twinship can be, it tends to
celebrate the twin relationship—elevating it to a unique stature in
the pantheon of close relationships. Although it is possible that
twin relationships are truly special and unlike any other kind of
human bond, we believe that the most appropriate way to establish
this possibility is by comparing systematically the features of twin
relationships with those of other kinds of close relationships.
Our results suggest that twin relationships are indeed special,
but they are not special in the sense that they represent a relation-
ship like no other. Instead, they appear to be attachment relation-
ships—relationships that, although exceptional, are experienced by
many people, twins and nontwins alike. There are several advan-
tages of conceptualizing twin relationships within the broader
framework of attachment theory. The most important advantage is
that attachment theory provides a general framework within which
to understand intimate relationships. By placing the study of twins
within a well-researched theoretical framework, it is possible to
refine the lens through which we view twin relationships. Attach-
ment theory provides a rich set of constructs, such as working
models, that might be valuable to study in twin relationships. It
would be useful to map the ways in which working models
contribute to dynamics that take place within the twinship, how
relational experiences might shape those representations, and how
those representations might be used (or modified) as twins forge
new relationships with romantic partners. It would also be inter-
esting, for example, to learn whether working models of the sibling
relationship, as opposed to the parental one, are more influential in
shaping the development of romantic relationships among twins as
opposed to nontwin siblings. These kinds of issues, although not
unique to attachment theory, flow naturally from an attachment
theoretical perspective on twin relationships. Although the present
research was focused on the normative aspects of attachment
theory rather than on individual differences, we believe that re-
Figure 2. A mediational model of the relations between genetic similarity, relational processes, and attach-
ment. Values represent standardized path coefficients. * ⫽ p ⬍ .05.
89
ATTACHMENT IN TWIN RELATIONSHIPS

search on individual differences in attachment organization would
be valuable for understanding the dynamics of twin relationships
(see Neyer, 2002a; Sheehan & Noller, 2002).
Implications for Attachment Theory
The majority of research on adult attachment has focused on
romantic relationships. Although this research has been fruitful, it
has often been conducted at the expense of studying other potential
attachment relationships, such as those between close friends (Fra-
ley & Davis, 1997) or between a person and his or her God
(Kirkpatrick, 1995). One of the implications of the present study is
that twinships constitute another important relationship that falls
within the province of attachment theory. In fact, there are several
ways in which twins might provide a more fertile testing ground
for basic attachment-theoretical concepts than that provided by
romantic relationships. Most important, twinships, unlike romantic
relationships, do not involve obvious elements of sexuality (except
in the minds, perhaps, of classical psychoanalysts). One of the
challenges in contemporary adult attachment research is to disen-
tangle the effects of mating strategies from the effects of attach-
ment security. For example, Kirkpatrick (1998) has argued that
what attachment researchers call “secure” versus “insecure” at-
tachment styles may reflect differences between long-term and
short-term mating strategies. At face value, Kirkpatrick’s (1998)
claim is credible. Secure adults, for example, tend to have longer
lasting relationships, experience less jealously, be less abusive, and
invest more in their offspring than insecure adults (Feeney, 1999).
Moreover, evolutionary psychologists have postulated that indi-
vidual differences in mating strategies have their origins in early
family relationships, with warm, supportive environments purport-
edly signaling the viability of long-term mating strategies and
unstable, low-resource environments signaling the viability of
short-term mating strategies (Belsky, 1999).
By studying attachment dynamics in intimate— but nonsexual
relationships—it should be possible to separate genuine attach-
ment phenomena from processes related to mating strategies. Of
course, the present research was not designed to accomplish this
task per se, but some of our findings are noteworthy in this regard.
For example, these data suggest that twin siblings are more likely
than nontwin siblings to rely upon one another as attachment
figures as they grow older. This pattern is not as strong in romantic
relationships. It is possible that the typical, relatively flat trajectory
observed for romantic relationships reflects two competing pro-
cesses. On one hand, the “natural” trajectory of an attachment
relationship may be to build on itself over time, as appears to be
the case with twins. On the other hand, romantic partners may be
keeping one ear to the ground, so to speak, seeking other potential
mates or guarding themselves emotionally against possible rejec-
tion or loss. They may also be noticing a decline in sexual
attraction or arousal within their relationship, which might make
them question whether the relationship is fulfilling all of their
needs. This kind of relational complexity may lead to what Fisher
(1992) has called the “4-year itch”—the modal breaking point of
married relationships across the globe—and to the bewildering
feelings of grief and loss following the breakup of a romantic/
sexual relationship that a person had thought was no longer satis-
fying or important (Weiss, 1975). In summary, the dynamics of
romantic relationships are thought to be influenced by both the
attachment behavioral system and the systems underlying mating
and reproduction (see Shaver, Hazan, & Bradshaw, 1988). These
systems are probably coordinated most of the time, making it
difficult to untangle their contribution to relational dynamics. By
focusing on long-lasting relationships that do not involve sexual-
ity, such as twin relationships, it is possible to study the way in
which attachment-related phenomena develop and impact people
in a manner that, at least theoretically, is free of sexuality.
10
Another feature of the twin relationship that might be beneficial
for understanding attachment relationships more broadly is that
many twins—in particular, monozygotic twins—are of the same
biological sex. Thus, the variation that typically exists in gender
roles and norms within heterosexual romantic relationships are
held constant in twin relationships. This would make it possible,
for example, to study caregiving dynamics between people who
are mutually attached without having those dynamics confounded
by the kinds of gender norms that exist in many cultures (Hupka &
Bank, 1996).
Attachment as a Mechanism of Inclusive Fitness
Despite the fact that attachment theory is ultimately a middle-
level evolutionary theory (Simpson, 1999), there has been some
tension between attachment theorists and evolutionary psycholo-
gists (Kirkpatrick, 1998). At first glance, a perspective that em-
phasizes the role of shared genes as a force that impacts emotional
bonding may seem incompatible with attachment theory—a theory
that has largely emphasized the role of people’s interpersonal
histories rather than their shared genes. One of the themes emerg-
ing from this research, however, is that these seemingly different
perspectives actually complement one another nicely. Specifically,
if one views relationship processes as mediators of the association
between genetic similarity and the development of attachment, it is
possible to discuss genes, interpersonal processes, and attachment
within a unified framework. In the research reported here, we
found that people were more likely to develop an attachment bond
with someone with whom they shared a larger proportion of their
genes. Our data also suggest that this association exists because
shared genes may affect relational dynamics, such as a heightened
sense of empathy, that promote the development of attachment.
Considering these three kinds of constructs as part of the same
causal chain, rather than as competing mechanisms, might provide
a useful bridge between evolutionary and social psychological
models of close relationships.
In summary, what has been missing from evolutionary and
genetic perspectives on social behavior is an explanation for how
genetic similarity is translated into something like an attachment
bond. Additionally, what has been missing from attachment theory
is an account of how evolutionary models of inclusive fitness
interface with what is known about attachment processes. Future
theoretical work— especially as it pertains to twin relationships—
would benefit from a better understanding of the complex inter-
connections between these different constructs (see Neyer & Lang,
2003, for a good illustration of how such work can be done).
10
This could also be accomplished by studying attachment dynamics in
friendships.
90
TANCREDY AND FRALEY

Limitations of the Present Research
Because this research is the first systematic investigation into
the nature of adult twin relationships from an attachment perspec-
tive, it is preliminary in many respects. The number of twins who
participated in the research was fairly small. It would be valuable
to continue this line of work with larger samples. A second
limitation of this research is that it does not focus on the twin
relationship from the perspective of both members of the dyad.
Like the majority of research on close relationships, this study
focuses on how individuals relate to others in their lives. Although
this information is valuable, there would be several advantages to
focusing on dyads rather than individuals. The most important
advantage would be that such a focus would allow more detailed
questions to be asked about the relationship. It would also allow
researchers to determine whether twins are more “in sync” with
one another than other pairs of siblings or relationship partners, as
is commonly reported. In an exemplary study along these lines,
Neyer (2002a) showed that there was an association between
feelings of security and dependency experienced among twins and
that both actor and partner effects played a role in shaping the
association between these variables among twin dyads and roman-
tic couples.
Third, one of the benefits of conceptualizing twin relationships
within the framework of attachment theory is the developmental
emphasis of the theory. Unfortunately, we were not able to study
attachment dynamics in a truly developmental fashion in this
cross-sectional research. Future research on these issues would
benefit enormously from adopting longitudinal designs. Finally,
our research has relied exclusively on self-report measures. Be-
yond the issues raised in the seemingly endless debates over
whether attachment can be assessed accurately via self-reports
(e.g., Crowell, Fraley, & Shaver, 1999), it would be valuable to
observe twin relationships in action and compare them with non-
twin relationships in domains such as nonverbal communication,
synchronicity (i.e., the “chameleon effect”; Chartrand & Bargh,
1999), proximity, and touch.
In closing, we note that the vast majority of psychological
research on twins has used them as tools—a means for quantifying
the contribution of genetic variation to human behavior. Although
the science of behavior genetics is obviously important for psy-
chology, the uniqueness of the twin situation calls for the study of
twin relationships in their own right. On the basis of the data
presented here, we can say that twin relationships are as close,
bonded, and nurturing as many romantic relationships and that it is
not typical for nontwin siblings to form this kind of bond. How-
ever, twin relationships are not in a special category in and of
themselves. They appear to belong to a broader class of relation-
ships—attachment relationships. It is our hope that an attachment-
theoretical perspective will allow us to understand twin relation-
ships better and, at the same time, advance our understanding of
intimate relationships more generally.
References
Ainslie, R. C. (1997). The psychology of twinship. Northvale, NJ: Jason
Aronson, Inc.
Ainsworth, M. D. S. (1989). Attachments beyond infancy. American Psy-
chologist, 44, 709–716.
Ainsworth, M. D. S. (1991). Attachments and other affectional bonds
across the life cycle. In C. M. Parkes, J. Stevenson-Hinde, & P. Marris
(Eds.), Attachment across the lifecycle (pp. 33–51). New York: Routledge.
Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns
of attachment: A psychological study of the strange situation. Hillsdale,
NJ: Erlbaum.
Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of other in the self
scale and the structure of interpersonal closeness. Journal of Personality
and Social Psychology, 63, 596 – 612.
Aron, A., Aron, E. N., Tudor, M., & Nelson, G. (1991). Close relationships
as including other in self. Journal of Personality and Social Psychology,
60, 241–253.
Bank, S. P., & Kahn, M. D. (1982). The sibling bond. New York: Basic
Books.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
51, 1173–1182.
Belsky, J. (1999). Modern evolutionary theory and patterns of attachment.
In J. Cassidy & P. Shaver (Eds.), Handbook of attachment: Theory,
research, and clinical applications (pp. 141–161). New York: Guilford.
Bouchard, T. J. (2004). Genetic influence on human psychological traits: A
survey. Current Directions in Psychological Science, 13, 148 –151.
Bowlby, J. (1969). Attachment and loss: Vol I. Attachment. New York:
Basic Books.
Burlingham, D. (1952). Twins: A study of three pairs of identical twins.
London: Imago.
Case, B. J. (1991). We are twins but who am I? Portland, OR: Tibbutt.
Cassidy, J. (1999). The nature of a child’s ties. In J. Cassidy & P. R. Shaver
(Eds.), Handbook of attachment: Theory, research, and clinical appli-
cations (pp. 3–20). New York: Guilford.
Chartrand, T. L., & Bargh, J. A. (1999). The chameleon effect: The
perception-behavior link and social interaction. Journal of Personality
and Social Psychology, 76, 893–910.
Collins, N. L., & Read, S. J. (1994). Cognitive representations of attach-
ment: The structure and function of working models. In K. Bar-
tholomew, & D. Perlman (Eds.), Attachment processes in adulthood (pp.
53–90). Bristol, PA: Jessica Kingsley.
Crowell, J., Fraley, R. C., & Shaver, P. R. (1999). Measures of individual
differences in adolescent and adult attachment. In J. Cassidy & P. R.
Shaver (Eds.), Handbook of attachment: Theory, research, and clinical
applications (pp. 434–465). New York: Guilford.
Doherty, N. A., & Feeney, J. A. (2004). The composition of attachment
networks throughout the adult years. Personal Relationships, 11, 469–
488.
Feeney, J. A. (1999). Adult romantic attachment and couple relationships.
In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory,
research, and clinical applications (pp. 355–377). New York: Guilford.
Fisher, H. (1992). The anatomy of love: The natural history of monogamy,
adultery, and divorce. New York: Norton.
Fraley, R. C., & Davis, K. E. (1997). Attachment formation and transfer in
young adults’ close friendships and romantic relationships. Personal
Relationships, 4, 131–144.
Hamilton, W. D. (1964). The genetical evolution of social behavior:I&II.
Journal of Theoretical Biology, 7, 1–52.
Hazan, C., Hutt, M. J., Sturgeon, J., & Bricker, T. (1991, April). The
process of relinquishing parents as attachment figures. Paper presented
at the biennial meetings of the Society for Research in Child Develop-
ment, Seattle, WA.
Hazan, C., & Shaver, P. R. (1987). Romantic love conceptualized as an
attachment process. Journal of Personality and Social Psychology, 52,
511–524.
Hazan, C., & Shaver, P. R. (1990). Love and work: An attachment-
theoretical perspective. Journal of Personality and Social Psychology,
59, 270 –280.
91
ATTACHMENT IN TWIN RELATIONSHIPS

Hazan, C., & Zeifman, D. (1994). Sex and the psychological tether. In K.
Bartholomew & D. Perlman (Eds.), Advances in personal relationships
(Vol. 5., pp. 151–180). London: Jessica Kingsley.
Heinicke, C. M., & Westheimer, I. (1965). Brief separations. New York:
International Universities Press.
Hinde, R. A. (1982). Ethology: Its nature and relations with other sciences.
Oxford, United Kingdom: Oxford Press.
Hupka, R. B., & Bank, A. L. (1996). Sex differences in jealousy: Evolution
or social construction? Cross-Cultural Research: The Journal of Com-
parative Social Science, 30, 24 –59.
Kirkpatrick, L. A. (1995). Attachment and religious experience. In R. W.
Hood, Jr., Handbook of religious experience: Theory and practice (pp.
446 – 475). Birmingham, AL: Religious Education Press.
Kirkpatrick, L. A. (1998). Evolution, pair-bonding, and reproductive strat-
egies: A reconceptualization of adult attachment. In J. A. Simpson &
W. S. Rholes (Eds.), Attachment theory and close relationships (pp.
353–393). New York: Guilford.
Kirkpatrick, L. A. (2004). Attachment, evolution, and the psychology of
religion. New York: Guilford.
Koch, H. L. (1966). Twins and twin relations. Chicago: University of
Chicago Press.
Lassers, E., & Nordan, R. (1978). Separation: Individuation of an identical
twin. Adolescent Psychiatry, 6, 469 – 479.
Leonard, M. (1961). Problems in identification and ego development in
twins. Psychoanalytic Study of the Child, 16, 300 –320.
Luria, A. R., & Yudovitch, F. I. (1959). Speech development of mental
processes in the child. London: Staple.
Lytton H. (1977). Parent– child interaction: The socialization process
observed in twin and singleton families. New York: Plenum.
Mahler, M. S., Pine, F., & Bergman, A. (1975). The psychological birth of
the human infant. New York: Basic Books.
Martin, R. W. (1991). Examining personal relationship thinking: The
relationship cognition complexity instrument. Journal of Social and
Personal Relationships, 8, 467–480.
Neyer, F. J. (2002a). The dyadic interdependence of attachment security
and dependency: A conceptual replication across older twin pairs and
younger couples. Journal of Social and Personal Relationships, 19,
483–503.
Neyer, F. J. (2002b). Twin relationships in old age: A developmental
perspective. Journal of Social and Personal Relationships, 19, 155–177.
Neyer, F. J., & Lang, F. R. (2003). Blood is thicker than water: Kinship
orientation across adulthood. Journal of Personality and Social Psychol-
ogy, 84, 310–321.
Peeters, H., Van Gestel, S., Vlietinck, R., Derom, C., & Derom, R. (1998).
Validation of a telephone zygosity questionnaire in twins of known
zygosity. Behavior Genetics, 28, 159 –161.
Sandbank, A. C. (1999). Personality, identity, and family relationships. In
A. C. Sandbank (Ed.), Twin and triplet psychology: A professional guide
to working with multiples (pp. 167–185). New York: Routledge.
Savic, S. (1980). How twins learn to talk. New York: Academic Press.
Schave, B., & Ciriello, J. (1983). Identity and intimacy in twins. New York:
Praeger.
Segal, N. L. (1997). Twin research perspective on human development. In
N. L. Segal, G. E. Weisfeld, & C. C. Weisfeld (Eds.), Uniting psychol-
ogy and biology: Integrative perspectives on human development (pp.
145–173). Washington, DC: American Psychological Association.
Segal, N. L. (1999). Entwined lives: Twins and what they tell us about
human behavior. New York: Dutton.
Segal, N. L., & Bouchard, T. J., Jr. (1993). Grief intensity following the
loss of a twin and other close relatives: Test of kinship-genetic hypoth-
esis. Human Biology, 65, 87–105.
Segal, N. L., & Ream, S. L. (1998). Decrease in grief intensity for deceased
twin and non-twin relatives: An evolutionary perspective. Personality
and Individual Differences, 25, 317–325.
Segal, N. L., Wilson, S. M., Bouchard, T. J., & Gitlin, D. G. (1995).
Comparative grief experiences of bereaved twins and other bereaved
relatives. Personality and Individual Differences, 18, 511–524.
Shaver, P. R., Hazan, C., & Bradshaw, D. (1988). Love as attachment: The
integration of three behavioral systems. In R. J. Sternberg & M. Barnes
(Eds.), The psychology of love (pp. 68 –99). New Haven, CT: Yale
University Press.
Sheehan, G., & Noller, P. (2002). Adolescent’s perceptions of differential
parenting: Links with attachment style and adolescent adjustment. Per-
sonal Relationships, 9, 173–190.
Simpson, J. A. (1999). Attachment theory in modern evolutionary perspec-
tive. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment:
Theory, research, and clinical applications (pp. 115–140). New York:
Guilford.
Sroufe, L., & Waters, E. (1977). Heart rate as a convergent measure in
clinical and developmental research. Merrill-Palmer Quarterly, 23,
3–27.
Stewart, R. B. (1983). Sibling attachment relationship: Child-infant inter-
actions in the strange situation. Developmental Psychology, 19, 192–
199.
Stewart, R. B., & Marvin, R. S. (1984). Sibling relations: The role of
conceptual perspective-taking in the ontogeny of sibling caregiving.
Child Development, 55, 1322–1332.
Tambs, K., Sundet, J. M., & Berg, K. (1985). Cotwin closeness in monozy-
gotic and dizygotic twins: A biasing factor in IQ heritability analysis?
Acta Geneticae Medicae et Gemellologiae: Twin Research, 34, 33–39.
Tancredy, C. M. (1999). Autonomy and connection in twin relationships.
Unpublished manuscript.
Tinbergen, N. (1963). On the aims and methods of ethology. Zeitschrift fu¨r
Tierpsychologie, 20, 410–433.
Trinke, S. J., & Bartholomew, K. (1997). Hierarchies of attachment rela-
tionships in young adulthood. Journal of Social and Personal Relation-
ships, 14, 603–625.
Vandell, D. L. (1990). Development in twins. Annals of Child Develop-
ment, 7, 145–174.
Weiss, R. S. (1975). Loneliness: The experience of emotional and social
isolation. Cambridge, MA: MIT Press.
Weiss, R. S. (1991). The attachment bond in childhood and adulthood. In
C. M. Parkes, J. Stevenson-Hinde, & P. Marris (Eds.), Attachment
across the lifecycle (pp. 66 –76). New York: Routledge.
Woodward, J. (1998). The lone twin: Understanding twin bereavement and
loss. London: Free Association Books.
92
TANCREDY AND FRALEY

Appendix
Attachment Features and Functions Measure
Instructions: Please take a moment to reflect on your current relationship
with _____ (mother, father, partner, sibling, twin, friend). If you had a
relationship with this individual, but he or she is now deceased or no longer
in your life, reflect on the relationship that existed when he or she was in
your life.
1. My _____ is the person that I would want to go to, to help me feel
better when something bad happens to me or I feel upset.
2. I make an effort to stay in contact with my _____.
3. If I achieved something good, my _____ is the person that I would tell
first.
4. My _____ is the person that I would like to be able to count on to
always be there for me and care about me no matter what.
5. My life would be severely disrupted if my _____ was no longer a part
of it.
6. My _____ is a person whom I count on for advice.
7. My _____ is the first person that I think of when I have a problem.
8. My _____ is the person that I would actually go to, to help me feel
better when something bad happens to me or I feel upset.
9. It is important to me to see or talk with _____ regularly.
10. My _____ is a person whom I do not like to be away from.
11. My _____ is the first person that I would turn to if I had a problem.
12. My _____’s death would have a great impact on me.
13. If my _____ was no longer accessible to me, I would feel greatly
distressed.
14. My ____ is my primary source of emotional support.
15. When I am away from my _____, I feel down.
16. My _____ is the person that I would actually count on to always be
there for me and care about me no matter what.
Note. Items 2 and 9 are indicators of proximity seeking. Items 1, 7, 8, 11,
and 14 are indicators of safe haven. Items 3, 4, 6, and 16 are indicators of
secure base. Items 5, 10, 12, 13, and 15 are items of separation distress.
Received October 28, 2004
Revision received March 31, 2005
Accepted April 13, 2005 䡲
93
ATTACHMENT IN TWIN RELATIONSHIPS