ArticlePDF Available

Abstract and Figures

Romantic couples (N = 194) participated in an investigation of caregiving processes in adulthood. In Phase 1, couple members completed questionnaires designed to identify attachment style differences in caregiving behavior and to explore the underlying (personal and relationship) mechanisms that lead people with different attachment styles to be effective or ineffective caregivers. Results revealed that social support knowledge, prosocial orientation. interdependence, trust, and egoistic motivation mediated the link between attachment style and caregiving. In Phase 2, responsive caregiving was assessed behaviorally by exposing one member of the couple to a stressful laboratory situation and experimentally manipulating his or her need for support. Results revealed that attachment style and mediating mechanisms identified in Phase 1 also predicted observable support behavior in a specific episode in which a partner had a clear need for support.
Content may be subject to copyright.
Journal
of
Personality
and
Social Psychology
2001.
Vol. 80, No. 6,
972-994
Copyright 2001
by the
American Psychological Association,
Inc.
OO22-3514/OI/S5.OO
DOI:
I0.1037//0022-3514.80.6.972
Predictors
of
Caregiving
in
Adult Intimate Relationships:
An Attachment Theoretical Perspective
Brooke C. Feeney
State University
of
New York
at
Buffalo
Nancy L. Collins
University
of
California, Santa Barbara
Romantic couples
(N =
194) participated
in an
investigation
of
caregiving processes
in
adulthood.
In
Phase
1,
couple members completed questionnaires designed
to
identify attachment style differences
in
caregiving behavior
and to
explore
the
underlying (personal
and
relationship) mechanisms that lead
people with different attachment styles
to be
effective
or
ineffective caregivers. Results revealed that
social support knowledge, prosocial orientation, interdependence, trust, and egoistic motivation mediated
the link between attachment style
and
caregiving.
In
Phase
2,
responsive caregiving
was
assessed
behaviorally by exposing one member
of
the
couple to
a
stressful laboratory situation and experimentally
manipulating
his or her
need
for
support. Results revealed that attachment style
and
mediating mecha-
nisms identified
in
Phase 1 also predicted observable support behavior
in a
specific episode
in
which
a
partner
had a
clear need
for
support.
Caregiving has been identified
as a
basic component
of
human
nature and
a
primary element
of
close relationships (Weiss, 1980).
Indeed, Bowlby (1988) argued that
the
capacity
to
make intimate
emotional bonds with others, sometimes
in the
care-seeking role
and sometimes
in the
caregiving role,
is a
principal feature
of
effective personality functioning
and
mental health.
It has
also
been argued that a healthy, well-functioning partnership is possible
only when the members are intuitively alive (Bowlby, 1988) to the
crucial role they play as caregivers and support providers (Brether-
ton, 1987).
In adulthood, romantic partners
are
frequently called
on to
provide comfort and assistance to one another in times
of
need,
and
many adults come
to
rely heavily
on
their romantic partner
as an
important
(if
not their most important) source
of
support and care.
It
is not
surprising that research indicates that social support
and
caregiving are essential not only
to
personal health
and
well-being
but also
to the
development
and
maintenance
of
healthy
and
satisfying intimate relationships (e.g., Acitelli, 1996; Carnelley,
Brooke C. Feeney, Department
of
Psychology, State University
of
New
York
at
Buffalo; Nancy
L.
Collins, Department
of
Psychology, University
of California, Santa Barbara.
Funding
for
this project was provided
by a
Mark Diamond Dissertation
Research Award (State University
of
New York
at
Buffalo). Preparation of
this article
was
supported
by
National Science Foundation Grant
SBR-
9870524.
We
gratefully acknowledge
the
contributions
of Joe
Chason,
Elaine Healy, Brent Jones, Kathy Morales, Alex Murray,
and
Tulin Ture,
who assisted with data collection and coding. Special thanks go to Carolyn
Cutrona, Paul Luce, Jack Meacham,
and
Sandra Murray
for
helpful com-
ments
and
suggestions
on
versions
of
this article.
Correspondence concerning this article should
be
addressed
to
Brooke
C. Feeney,
who is now at the
Department
of
Psychology, University
of
Maryland, College Park, Maryland 20742,
or to
Nancy L. Collins, Depart-
ment
of
Psychology, University
of
California, Santa Barbara, California
93106.
Electronic mail may
be
sent
to
brookefeeney@wam.umd.edu
or to
ncoiYms@psych.ucsb.edu.
Pietromonaco,
&
Jaffe, 1996; Collins
&
Feeney, 2000; Cutrona,
1996;
J. A.
Feeney, 1996; Julien
&
Markman, 1991). However,
despite
the
fact that caregiving
is
critical
for
well-being, we know
surprisingly little about the caregiving processes that occur in adult
close relationships. Most
of
the existing literature
has
focused
on
the experiences
of
the person who
is in
need
of
support, and much
less attention
has
been given
to the
person providing support.
There are many questions about caregiving that deserve research
attention. For example,
can
we identify individuals who are effec-
tive
and
ineffective caregivers?
If so,
what
are the
personal
and
interpersonal factors that explain these differences?
A
number
of
studies suggest that individual differences
in
adult attachment style
may
be one
important predictor
of
caregiving behavior
in
adult
close relationships. Thus,
the
current investigation takes
an in-
depth look
at
caregiving
in
adult intimate relationships from
an
attachment theoretical perspective. Specifically,
we
examine
at-
tachment style differences
in
caregiving
by
identifying specific
patterns
of
caregiving behavior
and by
exploring
a
number
of
possible mediators that
may
explain these differences.
We
inves-
tigate these issues
in a
large sample
of
adult romantic couples
using both self-report
and
laboratory methodologies.
Relevance
of
Attachment Theory
to the
Study
of Caregiving Processes
Attachment theory provides
an
ideal framework
for
studying
caregiving processes because
it
stipulates that the need
for
security
is a fundamental need
for
adults as well as
for
children and because
it provides
the
basis
for
understanding
the
complex attachment-
caregiving bond.
Attachment
According to attachment theory, individuals come into the world
equipped with
an
attachment behavioral system that
is
prone
to
activation when
an
individual
is
distressed,
and the
goal
of
this
972
PREDICTORS OF CAREGIVING
973
system
is to
maintain
a
feeling
of
security (Bowlby,
1973, 1980,
1982). Bowlby postulated that
the
attachment system serves
a
major evolutionary function
of
protection
and
survival;
it is
acti-
vated most strongly
in
adversity,
so
that when frightened, tired,
or
ill,
an
individual will seek protection
and
comfort from
a
primary
caregiver (Bowlby, 1982).
Because
not
everyone
has
experienced responsive
and
reliable
caregiving
in
times
of
need,
not
everyone expects
to get
their
attachment needs
met.
This
has led
researchers
to
identify individ-
ual differences
in
attachment styles
or
patterns (Ainsworth, Blehar,
Waters,
&
Wall, 1978; Bartholomew
&
Horowitz, 1991; Hazan
&
Shaver, 1987). Attachment styles
can be
thought
of as
chronic
interpersonal styles that reflect people's general beliefs about
themselves
and
others—beliefs about whether
the
self
is
worthy
of
care
and
affection,
and
beliefs about whether other people
are
generally dependable
and
responsive.
The
different patterns
of
attachment also represent different strategies
of
affect regulation,
or rules that guide individuals' responses
to
emotionally distress-
ing situations (Kobak
&
Sceery, 1988).
Adult attachment researchers typically define four prototypic
attachment styles derived from
two
underlying dimensions—anx-
iety
and
avoidance (e.g., Bartholomew
&
Horowitz,
1991;
Bren-
nan, Clark,
&
Shaver, 1998).
The
anxiety dimension assesses
the
degree
to
which
the
self
is
perceived
to be
worthy
or
unworthy
of
love
and the
degree
to
which
the
individual
is
worried about being
rejected
by
others. Anxious attachment
is
organized
by
rules that
direct attention toward distress
and
attachment figures
in a
hyper-
vigilant manner, which inhibits
the
development
of
autonomy
and
self-confidence.
The
avoidance dimension assesses
the
degree
to
which individuals perceive others
to be
generally responsive
or
unresponsive
and the
degree
to
which individuals
are
comfortable
with intimacy
and
dependency
on
others. Avoidant attachment
is
organized
by
rules that restrict acknowledgment
of
distress
as
well
as
any
attempts
to
seek comfort
and
support from others.
The
four
attachment styles
can be
conceptualized
in
terms
of
these
two
underlying dimensions
as
follows; Secure adults
are low in
both
anxiety
and
avoidance, preoccupied adults
are
high
in
anxiety
and
low
in
avoidance, fearful avoidant individuals
are
high
in
both
anxiety
and
avoidance,
and
dismissing avoidant individuals
are
low
in
anxiety
but
high
in
avoidance.
Caregiving
Attachment theory stipulates that
the
caregiving system
is an-
other normative, safety-regulating system that
is
intended
to re-
duce
the
risk
of a
close other coming
to
harm. Caregiving
has
been
defined
as a
broad array
of
behaviors that complement
a
partner's
attachment behavior
and may
include
a
wide range
of
responsibil-
ities,
such
as
providing help
or
assistance, offering comfort
and
reassurance, providing
a
secure base,
and
encouraging autonomy
(Bowlby,
1982, 1988;
Kunce
&
Shaver, 1994). Bowlby (1982,
1988) identified
the
need
to
study
the
caregiving system within
a
conceptual framework similar
to
that adopted
for
attachment
be-
havior. However, with only
a few
exceptions (e.g., Carnelley
et al.,
1996; Collins
&
Feeney, 2000;
J. A.
Feeney,
1996;
George
&
Solomon,
1996,
1999; Kunce
&
Shaver, 1994; Solomon
&
George,
1996),
the
study
of
caregiving
as a
behavioral system
has
received
little attention.
Attachment Style Differences in Caregiving
Because working models
of
attachment (attachment styles)
are
built interactionally
and
encompass views
of
both self
and
others
(both sides
of the
attachment-caregiving relationship), beliefs
about
the
likelihood
of
receiving care from others
(and
rules that
guide support-seeking behavior
and the
regulation
of
personal
distress)
are
likely
to be
linked
to
beliefs about providing care
to
others
(and
rules that guide caregiving behavior
and the
regulation
of
a
significant other's distress; Collins
&
Feeney, 2000; Kunce
&
Shaver, 1994). Kunce
and
Shaver (1994) were
the
first
to
identify
the links between attachment style
and
caregiving patterns
in
adult
intimate relationships,
and
other researchers have shown
(in self-
report
and
observational studies) that attachment style
is
system-
atically associated with caregiving. We briefly review these studies
below.
Self-Reported
Caregiving
Behaviors
In
a
series
of
self-report studies, attachment researchers have
shown that each attachment style
is
associated with
a
unique
pattern
of
caregiving (Carnelley
et al., 1996; J. A.
Feeney,
1996;
Kunce
&
Shaver, 1994). Specifically, secure individuals
(low
anxiety
and low
avoidance) report relatively high levels
of
sensi-
tivity
and
proximity (i.e., physical forms
of
comfort), coupled with
relatively
low
levels
of
compulsive (i.e., overinvolved)
and con-
trolling caregiving.
In
contrast, preoccupied individuals (high
anx-
iety
and low
avoidance) report relatively
low
levels
of
sensitivity
and cooperation
but
relatively high levels
of
proximity
and com-
pulsive caregiving—suggesting that their caregiving behaviors
may
be
inconsistent, intrusive,
and out of
sync with their partner's
needs. Dismissing individuals
(low
anxiety
and
high avoidance)
report
the
lowest levels
of
compulsive caregiving
and
provision
of
proximity,
and
they also report relatively
low
levels
of
sensitivity.
Finally, fearful individuals (high anxiety
and
high avoidance)
report relatively
low
levels
of
sensitivity
and
proximity, while
simultaneously reporting relatively high levels
of
compulsive
caregiving.
Observed
Caregiving
Behaviors
Although observational research
is
still relatively rare
in the
social support literature, several attachment researchers have
ex-
amined whether self-reported attachment style
is
predictive
of
directly observable caregiving behavior. Simpson, Rholes,
and
Nelligan (1992) observed
the
spontaneous caregiving behavior
offered
by
male caregivers when
a
female dating partner
was
exposed
to an
anxiety-provoking experimental procedure. Secure
men offered more comfort
and
reassurance, whereas avoidant
men
were less inclined
to
offer support
(and
displayed more anger)
as
their partners displayed greater anxiety
(see
also Rholes, Simpson,
& Orina, 1999). However,
no
significant effects emerged
for
caregivers
who had an
anxious attachment style. This pattern
of
results
was
corroborated
in a
field study conducted
by
Fraley
and
Shaver (1998),
in
which they observed couples separating
at an
airport.
In a
more recent study
in
which participants were video-
taped
as
they discussed
a
current stressful event
in
their lives with
their romantic partners, Collins
and
Feeney (2000) found that
caregivers
who
were high
in
anxiety were less responsive during
974
FEENEY AND COLLINS
the interaction and provided less positive (e.g., emotional and
instrumental support) and more negative (e.g., dismissing the
problem, blaming the partner) forms of support. In another study,
in which romantic couples were unobtrusively videotaped after
one member of the couple was asked to give a speech, B. C.
Feeney and Collins (1998) found that caregiver avoidance was
associated with less responsiveness, less proximity seeking, and
less warmth and friendliness. Important links between adult at-
tachment representations and caregiving behaviors have also been
shown in several observational studies of mothers and children
(Crowell & Feldman, 1988, 1991; see van IJzendoorn, 1995, for a
meta-analytic review). In all of these studies, secure adult attach-
ment was associated with sensitive, warm, and positive maternal
caregiving behaviors.
Why
Are
People With Different Attachment Styles
Effective
or
Ineffective Caregivers?
On the basis of the studies reviewed above, it appears that
individuals with a secure attachment style (low anxiety, low avoid-
ance) tend to be good, responsive caregivers: They are warm,
sensitive, and cooperative, and they actively help their partners
solve problems. Insecure adults, on the other hand, tend to be
relatively poor caregivers, but they exhibit different forms of
unresponsive caregiving depending on their particular type of
insecurity. It is important to note, however, that the few observa-
tional studies that have examined links between attachment style
and behavioral measures of caregiving provided to romantic part-
ners to date have yielded somewhat inconsistent results, most
likely as a result of differences in the particular laboratory situation
or support context. That is, in two studies in which participants
were unobtrusively observed in stressful situations, avoidant indi-
viduals emerged as particularly poor caregivers (Simpson et al.,
1992;
B. C. Feeney & Collins, 1998); however, another study, in
which couples discussed a problem identified as stressful to one
member of the couple, revealed that anxious individuals were the
especially poor and ineffective caregivers (Collins & Feeney,
2000).
Although researchers have shown that attachment style is sys-
tematically associated with caregiving behavior, we currently
know little about the specific mediating mechanisms that can
explain these differences. Why are people with different attach-
ment styles effective or ineffective caregivers? Discovering the
answer to this question is important for several reasons. First, it
enables us to gain greater insight into individual differences in
attachment style and greater understanding of the interface be-
tween the attachment and caregiving systems. Second, it enables
us to learn, more broadly, about the particular skills, resources,
and motivations that are associated with caregiving in close
relationships.
What are the necessary ingredients for caregiving? For one to
identify the mediators that explain attachment style differences in
caregiving, it is important to consider the factors that are required
for effective caregiving. We believe that three major ingredients
are necessary for the provision of responsive care and support and
that insecure individuals may be relatively poor caregivers because
they lack one or more of these necessary ingredients.
First, effective caregiving requires individuals to possess rele-
vant skills and abilities. For example, individuals must be able to
respond flexibly to a wide range of needs as they arise, and
caregivers must have adequate knowledge about how to provide
the appropriate type and amount of support that is needed. Addi-
tional skills important for sensitive caregiving include (a) the
ability to empathize with and take the perspective of distressed
individuals and (b) social skills that assist caregivers in orienting
themselves toward others and recognizing their needs.
Second, effective caregiving requires adequate emotional and
material resources. In describing the conditions necessary for
parents to be sensitive and attentive caregivers, Bowlby (1988)
acknowledged that adequate time and a relaxed atmosphere are
necessary. These same resources should be necessary for adults to
be responsive to one another. For example, if an individual is
stressed, overwhelmed with work or family responsibilities, and
experiencing time constraints, it is likely that his or her caregiving
behavior will suffer because he or she will be self-focused and may
temporarily lack the emotional energy and cognitive resources
necessary to devote to his or her partner. Because the attachment
and caregiving roles are not exclusively assigned to one member of
a dyad in adult relationships, adult partners may frequently expe-
rience stressful events concurrently, and recent studies suggest that
caregiving quality may deteriorate under these circumstances
(Vinokur, Price, & Caplan, 1996; Wood, Saltzberg, & Goldsamt,
1990).
It is important to note that one can lack resources either
chronically (e.g., chronic self-focus) or situationally (e.g.,
situation-specific self-focus).
Finally, an individual must have the motivation to provide care.
The caregiving role often involves a good deal of responsibility as
well as a substantial amount of cognitive, emotional, and some-
times tangible resources. Therefore, caregivers must be motivated
to accept that responsibility (which often involves some degree of
sacrifice) and expend the time and effort required to provide
effective support. If caregivers are not sufficiently motivated, they
may provide either low levels of care or ineffective forms of
caregiving that are out of sync with their partner's needs. Motiva-
tions for caregiving may be chronic (e.g., a general communal
orientation toward others), relationship-specific (e.g., a feeling of
commitment to and responsibility for a particular individual), or
situation-specific (e.g., feeling pressured or obligated to provide
support in a particular situation).
The Current Investigation
The current investigation focuses explicitly on the caregiver to
address two specific research goals. The first goal is to provide
further evidence for attachment style differences in caregiving
behavior by using self-report and laboratory methods and by
exploring a wider variety of caregiving patterns than have been
examined in prior research. The second goal is to provide a more
detailed understanding of caregiving dynamics by identifying the
mediators of these attachment style differences in caregiving.
Specifically, we seek to identify the personal and interpersonal
factors (on the basis of the ingredients for good caregiving listed
above) that may lead people with different attachment styles to be
effective or ineffective caregivers. Our interest is both in explain-
ing attachment style differences in caregiving and, more generally,
in understanding the predictors of effective caregiving.
We addressed these objectives in a two-phase investigation
involving romantic couples. We used a self-report methodology
PREDICTORS
OF
CAREGIVING
975
during
the
first phase
to
identify
the
factors that predict various
forms
of
caregiving
and
lead individuals with different attachment
styles
to be
effective
or
ineffective caregivers.
In the
second phase
of this investigation,
we
used
an
experimental-observational
methodology
to
obtain behavioral evidence
of
caregivers' respon-
siveness
and
attentiveness
to
their partner's needs.
Phase
1
(Questionnaire Session): Hypotheses
Attachment Style Differences in Patterns of
Caregiving
Before
we
examine
the
mediational hypotheses,
it is
important
to replicate
and
extend
the
associations between attachment style
and caregiving that have been observed
in
previous studies.
Con-
sistent with recent theoretical
and
empirical advances
in the
field
(Brennan
et al., 1998;
Fraley
&
Waller, 1998),
our
hypotheses
concerning attachment style differences focus
on the
dimensions
of attachment-related anxiety
and
avoidance.
On the
basis
of
theory
and
previous research,
we
expected that avoidance would
be associated with
an
unfavorable caregiving style characterized
by
low
levels
of
responsive caregiving
and
high levels
of
control-
ling caregiving.
In
contrast,
we
expected that attachment-related
anxiety would
be
associated with
a
caregiving style characterized
by high levels
of
both compulsive
and
controlling caregiving.
Because anxiety
has
been inconsistently associated with unrespon-
sive caregiving
in
prior studies, this relationship
was
more difficult
to predict.
On the one
hand, anxious adults
are
worried about being
rejected
and
unloved
by
others,
and
they tend
to
direct attention
toward attachment figures
in a
hypervigilant manner; therefore,
they
may
respond
to
their partner's needs with emotional support
and physical proximity. However, their fear
of
abandonment
and
lack
of
confidence
in the
partner's continuing love
and
commit-
ment
may
impede their ability
to
provide responsive support that
is
in sync with
the
needs
of the
partner.
Proposed Mediators of
the
Link Between Attachment
Style and
Caregiving
Quality
The second
set of
hypotheses concerns
the
identification
of the
particular types
of
skills, resources,
and
motivations that
are
likely
to influence caregiving quality
and
mediate
the
link between
attachment style
and
caregiving.
We
organized these mediators
into personal/individual-level factors
and
relationship-level factors
(see Figure
1).
Personal/Individual-Level
Factors
On the basis of the available literature on close relationships
as
well
as
the
social support
and
helping literatures,
we
identified four rela-
tively chronic individual-level factors that
may
predict caregiving
quality and mediate the link between attachment style and caregiving:
(a) empathy
(an
interpersonal skill
or
ability),
(b)
social support
knowledge
(a
skill
or
ability),
(c)
chronic
self-focus
(lack
of a
cogni-
tive resource),
and (d)
communal
and
exchange orientation (general
interpersonal motivations). First, numerous studies
in the
helping
literature have indicated that under conditions of empathic concern
for
another, individuals help more frequently
in
what appears
to be an
altruistically motivated attempt
to
improve
the
other's well-being
(e.g.,
Batson, O'Quin, Fultz, Vanderplas,
&
Isen,
1983;
Dovidio,
Allen,
&
Schroeder, 1990). Therefore,
we
expected that empathic
abilities would
be
associated with more responsive forms
of
caregiv-
ing.
Second, knowledge about effective supportive behavior should
ATTACHMENT
STYLE
PERSONAL/INDIVIDUAL LEVEL MEDIATORS
Empathic Abilities
Social Support Knowledge/Efficacy Beliefs
Chronic Self-Focus
Communal and Exchange Orientation
RELATIONSHIP LEVEL MEDIATORS
Commitment Variables:
Commitment
Investment in Relationship
Quality of Alternatives
Closeness Variables:
Subjective Closeness
Interdependence
Trust
Specific Motivations for Caregiving
Altruistic Motives
Egoistic Motives
CAREGIVING
QUALITY
Figure 1. Summary of variables hypothesized to mediate the link between attachment style and caregiving.
976
FEENEY AND COLLINS
increase the actual support behavior that one romantic partner offers
to the other (Johnson, Hobfoll, & Zalcberg-Linetzy, 1993). Third,
because cognitive resources are needed to monitor a partner's needs
and to determine how to respond appropriately, chronically
self-
focused individuals are likely to provide low levels of care (because
they fail to notice their partner's needs) or ineffective care that is out
of sync with their partner's needs. And finally, because communal
and exchange orientation are interpersonal proclivities that involve
beliefs about the giving and receiving of benefits (Mills & Clark,
1994),
a communal orientation toward others should be associated
with sensitive caregiving, whereas an exchange orientation should be
associated with less effective forms of caregiving (Clark, Ouellette,
Powell, & Milberg, 1987).
Relationship-Level
Factors
We also reasoned that four categories of relationship-level factors
may predict caregiving quality and mediate the link between attach-
ment style and caregiving: (a) commitment, (b) closeness and inter-
dependence, (c) trust, and (d) altruistic versus egoistic motives for
caring for one's partner. We expected all of these relationship factors
to function as relationship-specific motivating forces in eliciting care-
giving behavior. First, relationship commitment provides individuals
with the motivation to engage in pro-relationship behaviors (behav-
iors that benefit the relationship and help it survive)—even when such
behaviors are costly or stand in direct opposition to self-interest (e.g.,
Rusbult & Buunk, 1993). We expected that responsive caregiving
would be an important pro-relationship behavior. Second, feelings of
subjective closeness to one's partner should lead individuals to be
willing to give benefits in response to need (e.g., Clark, Mills, &
Powell, 1986; Clark et al., 1987) because the partners' lives have become
deeply intertwined and the boundary between self-interest and partner
interest is blurred (e.g., Aron, Aron, & Smollan, 1992). Good caregiving
should be observed in close, interdependent relationships because indi-
viduals feel responsible for the well-being of their partners and are
motivated to improve the joint welfare of the relationship.
Third, because trust involves confidence in a partner's responsive-
ness to one's needs as well as a willingness to put oneself at risk
(Holmes, 1991), individuals who trust their partners should be good
caregivers for two reasons: Individuals who are confident that a
partner will be responsive to their needs should be more likely to
strive to meet their partner's needs, and individuals who are willing to
put themselves at risk should be better caregivers because caregiving
often involves some risk or cost to
oneself.
Finally, we reasoned that
relationship-specific egoistic and altruistic motivations influence the
quality of care (e.g., Batson et al., 1991; Cialdini, Schaller, Houlihan,
Arps,
Fultz, & Beaman, 1987). Individuals who are egoistically mo-
tivated to care for their partners (to receive rewards or avoid negative
consequences) may be poor caregivers because they provide the type
of support that is more beneficial to themselves than to the partner. In
contrast, individuals who are relatively altruistically motivated (out of
genuine concern for the partner's well-being) may be good caregivers
because they provide support that is dictated by the partner's needs.
Mediation Hypotheses
On the basis of the attachment literature and the expected patterns
of caregiving, we proposed the following mediational hypotheses.
According to the attachment literature, avoidant individuals tend to
direct their attention away from attachment needs, emphasize inde-
pendence and self-reliance, and tend to be involved in relationships
characterized by low levels of commitment, interdependence, and
trust (Bartholomew & Horowitz, 1991; Collins & Read, 1990; J. A.
Feeney & Noller, 1990; Hazan & Shaver, 1987; Kobak & Sceery,
1988;
Simpson, 1990). Therefore, we expected avoidant individuals
to be unresponsive and controlling caregivers because of their low
levels of empathy, communal orientation, support knowledge, com-
mitment, relationship closeness, trust, and altruistic caregiving moti-
vation. We also expected avoidant individuals to be unresponsive and
controlling caregivers because of their high levels of self-focus, ex-
change orientation, and egoistic caregiving motivation.
The attachment literature indicates that anxious individuals fear
being abandoned and unloved, tend to be dependent on others'
acceptance of them for a sense of personal well-being, direct
attention toward attachment figures in a hypervigilant manner, and
express distrusting views of others (Bartholomew & Horowitz,
1991;
Collins & Read, 1990; J. A. Feeney & Noller, 1990; Hazan
& Shaver, 1987; Kobak & Sceery, 1988; Simpson, 1990). There-
fore,
we expected anxious individuals to be relatively compulsive
and controlling caregivers and perhaps also unresponsive because
of their low levels of trust and altruistic caregiving motivation and
their high levels of self-focus, exchange orientation, and egoistic
caregiving motivation. It is also possible that the high levels of
relationship commitment and subjective closeness (characteristic
of anxious individuals) may contribute to the link between anxiety
and compulsive caregiving.
Method
Overview
In this phase of the project, we used questionnaire methodology to assess
three sets of factors necessary for testing the hypotheses outlined above: (a)
individual differences in the dimensions of attachment style (avoidance and
anxiety), (b) a variety of personal- and relationship-level mediator vari-
ables,
and (c) several comprehensive assessments of caregiving.
Participants
Participants were 202 couples from the State University of New York at
Buffalo and the University of California, Santa Barbara. One member of
each couple was recruited from the introductory psychology participant
pool and was asked to bring his or her romantic partner to the study. For
each phase of this study, the recruited introductory psychology student was
designated as the "support recipient," and his or her romantic partner was
designated as the "caregiver."
1
The participant of interest was the person in
the role of caregiver. The mean age of support recipients was 19.1 (range =
17-33),
and the mean age of caregivers was 19.5 (range = 17-28). Couples
1
For purposes of this investigation, it was necessary to focus on one
member of each couple as
a
caregiver and on the other member as a support
recipient. This assignment of roles was dictated by Phase 2 of the project.
To ensure the validity of responses from all participants during Phase 2, it
was important that the undergraduate psychology participant be designated
as the support recipient (the person who would receive the stress induction
in Phase 2) and that his or her partner be designated as the caregiver (the
person who would not undergo the stress induction). Because romantic
partners often participate in our studies as a favor to the undergraduate
psychology participants, there was a danger that the validity of the ob-
served caregiving behaviors in Phase 2 would be compromised if the roles
PREDICTORS OF CAREGIVING
977
had been romantically involved for an average of 14.4 months (range =
1-95),
and all were heterosexual. The majority of couples were involved in
dating relationships (93%), and a small percentage were either married or
engaged to be married (7%).
Couple members either received course credit for their participation or
were paid $10 and offered an opportunity to win a $100 prize in a drawing.
Of the 202 original couples, 8 couples were excluded from data analyses,
either because they were not proficient in English or because their involve-
ment in an established romantic relationship was questionable. Of the
remaining 194 couples, 111 men and 83 women were assigned to the
caregiver role. In analyses involving the attachment dimensions, 2 addi-
tional couples were excluded because they did not complete the attachment
Procedure
Couples were informed that they would be participating in a two-part
investigation that was designed to help researchers learn more about the
ways people think, feel, and behave in their relationships with their
romantic partners. During Phase 1, couple members completed question-
naires in separate, private rooms. An appointment for Phase 2 was then
scheduled for approximately 1 week later. The measures used in Phase 1
are described below.
Measures
Couples completed two attachment measures. First, they completed
Brennan et al.'s (1998) 36-item attachment scale, which contains two
subscales: The Avoidance subscale (a = .92) measures the extent to which
a person is comfortable with closeness and intimacy as well as the degree
to which a person feels that people can be relied on to be available when
needed. The Anxiety subscale (a = .92) measures the extent to which a
person is worried about being rejected, abandoned, or unloved. Couples
responded to each item on a scale ranging from 1
(strongly disagree)
to 5
(strongly
agree) in terms of their general orientation toward close relation-
ships.
Next, couples were presented with Bartholomew and Horowitz's
(1991) four attachment prototypes (secure, fearful, preoccupied, dismiss-
ing) and were asked to rate the extent to which each one corresponded to
their general style in romantic relationships. We computed an avoidance
dimension by subtracting the secure and preoccupied ratings from the sum
of the dismissing and fearful ratings, and we computed an anxiety dimen-
sion by subtracting the secure and dismissing ratings from the sum of the
preoccupied and fearful ratings.
To achieve the most valid and reliable assessment of attachment style,
we combined these measures to form two composite attachment dimen-
sions.
The two avoidance indices were highly correlated (r = .67, p <
.001) and were combined to form a composite avoidance dimension. High
scores on this dimension reflect discomfort with closeness and a tendency
to avoid intimate relationships. Likewise, the two anxiety indices were
highly correlated (r = .64, p < .001) and were combined to create a
composite anxiety dimension. High scores reflect a sense of low self-worth
and anxiety about being rejected by others. The avoidance and anxiety
composites were not significantly correlated with each other (r = .08, ns).
Personal/Individual-Level
Mediating Variables
Empathy
and
perspective
taking.
To obtain the most valid and reliable
assessment of dispositional empathy, we had couples complete a 20-item
were reversed (i.e., if the person who was participating as a favor to his or
her partner received the stress induction). However, if both members of the
couple were recruited from the undergraduate participant pool, then they
were either randomly assigned to roles or assigned on the basis of their
attachment characteristics (to ensure an appropriate number of insecure
individuals in the caregiving role).
measure derived from two widely used empathy scales: the Empathi