Creating and Undermining Social Support

INTERPERSONAL RELATIONS AND GROUP PROCESSES
Creating and Undermining Social Support in Communal Relationships:

The Role of Compassionate and Self-Image Goals
Jennifer Crocker and Amy Canevello
University of Michigan
In 2 studies, the authors examined whether relationship goals predict change in social support and trust
over time. In Study 1, a group of 199 college freshmen completed pretest and posttest measures of social
support and interpersonal trust and completed 10 weekly reports of friendship goals and relationship
experiences. Average compassionate goals predicted closeness, clear and connected feelings, and
increased social support and trust over the semester; self-image goals attenuated these effects. Average
self-image goals predicted conflict, loneliness, and afraid and confused feelings; compassionate goals
attenuated these effects. Changes in weekly goals predicted changes in goal-related affect, closeness,
loneliness, conflict, and beliefs about mutual and individualistic caring. In Study 2, a group of 65
roommate pairs completed 21 daily reports of their goals for their roommate relationship. Actors’ average
compassionate and self-image goals interacted to predict changes over 3 weeks in partners’ reports of
social support received from and given to actors; support that partners gave to actors, in turn, predicted
changes in actors’ perceived available support, indicating that people with compassionate goals create a
supportive environment for themselves and others, but only if they do not have self-image goals.
Keywords: social support, goals, self-image, compassion, relationships
Humans are social animals; they need supportive relationships likes (or loves) them (Baumeister & Leary, 1995). In support of
with other people for physical and psychological well-being (e.g., their argument, Baumeister and Leary (1995) reviewed evidence
Bowlby, 1969; Deci & Ryan, 2000; Maslow, 1968). Baumeister that people form social bonds easily and are reluctant to break
and Leary (1995) proposed that humans have a fundamental need them; that forming social bonds creates positive emotions, whereas
to belong; specifically, people need frequent personal interaction breaking social bonds creates negative emotions; that people think
or contact with someone who cares about their welfare and who a great deal about actual and potential relationship partners; and
that deficits in belongingness are associated with both physical and
mental health problems. Fiske (2003) suggested that belonging is
the core social motive in humans, underlying the motives to
Jennifer Crocker, Research Center for Group Dynamics, Institute for
understand, control, self-enhance, and trust in others. Consistent
Social Research, and Department of Psychology, University of Michigan;
Amy Canevello, Institute for Social Research, University of Michigan. with this view, social support predicts both physical and psycho-
The project described was supported by Grant R01MH058869 from the logical health (Cohen & Syme, 1985; Pierce, Sarason, & Sarason,
National Institute of Mental Health. The content is solely the responsibility 1996; B. R. Sarason, Sarason, & Gurung, 1997; Uchino, 2004;
of Jennifer Crocker and Amy Canevello and does not necessarily represent Uchino, Cacioppo, & Kiecolt-Glaser, 1996). Furthermore, the per-
the official views of the National Institute of Mental Health or the National ception that others are available and supportive predicts health and
Institutes of Health. well-being better than does objective social support received from
We are grateful to Rebecca Caulfield, Timothy Cavnar, Andrew
others (Cohen & Syme, 1985).
Crocker, Sarah Franz, and Riia Luhtanen for their assistance with Study 1
and to Juliana G. Breines and Mary Y. Liu for their assistance with Study Although social support can buffer the effects of negative life
2. We are also grateful to Brooke Feeney and Linda Roberts for their events (Burton, Stice, & Seeley, 2004; Cropley & Steptoe, 2005;
helpful comments on a previous version of this article and to our friends at Uchino, 2004), life events can disrupt available social support,
Learning as Leadership for inspiring our research on compassionate and with consequences for physical and psychological well-being.
self-image goals. Some disruptions of available social support, such as the death of
Correspondence concerning this article should be addressed to Jennifer
a spouse, are uncontrollable; other disruptions may be side effects
Crocker, Research Center for Group Dynamics, Institute for Social Re-
search, University of Michigan, 426 Thompson Street, Ann Arbor, Mich-
of otherwise positive life events, such as moving away from home
igan 48106, or to Amy Canevello, Institute for Social Research, University to attend college or to take a job. When such disruptions occur, the
of Michigan, 426 Thompson Street, Ann Arbor, Michigan 48106. E-mail: capacity to rebuild available social support may influence whether
jcrocker@umich.edu or canevell@umich.edu people adjust to the event or transition with good mental and
Journal of Personality and Social Psychology, 2008, Vol. 95, No. 3, 555–575
Copyright 2008 by the American Psychological Association 0022-3514/08/$12.00 DOI: 10.1037/0022-3514.95.3.555
555
556 CROCKER AND CANEVELLO
physical health or succumb to mental and physical health prob- competing to maximize their relative advantage over others, are
lems. said to have a prosocial value orientation (Van Lange et al., 2002).
Supportive interactions involve a support provider, a support Alternatively, people may act supportively not because they
recipient, and the situation, each of which contributes to the expect direct reciprocation but because they genuinely care about
effectiveness of a support transaction (I. G. Sarason, Pierce, & the welfare of others and because the needs of others are salient
Sarason, 1990). Characteristics of support recipients, such as self- (Batson, 1998; Brown & Brown, 2006; Collins & Feeney, 2000;
esteem, depressive cognitive styles, and attachment styles, corre- B. C. Feeney & Collins, 2003). Giving in response to a partner’s
late with perceived available social support and shape interpreta- needs builds close relationships (see Reis, Clark, & Holmes, 2004,
tions of supportive behaviors (Collins & Feeney, 2004; Lakey & for a review). For example, attachment theory suggests that secure
Cassady, 1990). Characteristics of support providers also shape attachment bonds develop when others are reliably sensitive and
judgments of support (Lakey, McCabe, Fisicaro, & Drew, 1996). responsive to one’s needs (Bowlby, 1969). In adulthood, relation-
For example, providers’ agreeableness influences perceivers’ judg- ship satisfaction in dating and married couples depends on per-
ments of support (Lakey et al., 2002; Lakey, Drew, & Sirl, 1999; ceiving that one’s partner is a good caregiver who provides a safe
Lutz & Lakey, 2001). Features of the situation, such as the events haven of comfort and security (Carnelley, Pietromonaco, & Jaffe,
that create the need for support and the type of relationship 1996; J. A. Feeney, 1996). When people experiencing distress
between the provider and the recipient also influence support perceive their partner’s responses as supportive, their mood im-
transactions (Lakey et al., 1996; I. G. Sarason et al., 1990). proves (Collins & Feeney, 2000). Caregiving associated with a
Furthermore, characteristics of the provider, the recipient, and the prosocial orientation (e.g., “I love my partner and am concerned
situation may interact to influence judgments of support (Lakey et about my partner’s well-being”) positively predicts how respon-
al., 1996). Thus, factors that influence judgments of the support- sive care giving is in intimate relationships and how supported
iveness of a single support transaction are increasingly well un- intimate partners feel under conditions of need (B. C. Feeney &
derstood. Collins, 2001). People feel close and connected to others whom
Less is known, however, about the dynamics of social support they perceive to be responsive and supportive (LeMay, Clark, &
over time. Do some people effectively build support for them- Feeney, 2007). Feelings of closeness, in turn, foster the desire to
selves when their support networks have been disrupted? Do some give support to others (Brown & Brown, 2006).
people undermine their social support over time? If so, what These findings suggest that people who give support in response
differentiates those people who experience increases in support to other’s needs and out of concern for other’s welfare build close
from those who do not? What intrapersonal and interpersonal relationships in which they ultimately receive support, even though
processes characterize upward trajectories of social support? obtaining support is not their goal. Relationships governed by
concern for others’ welfare, in which benefits are given in response
to others’ needs or to please others are called communal relation-
Building Support by Being Supportive ships (Clark et al., 1986, p. 333). People who dispositionally tend
to prioritize responsiveness to needs are said to have a communal
Several theoretical perspectives suggest that people can build relationship orientation (Clark, Oullette, Powell, & Milberg, 1987;
supportive relationships, and consequently increase the support Mills, Clark, Erber, & Gilmour, 1994).
available to them, by giving support to others. Evolutionary theory, We propose that the emotional bond the support provider feels
game theory, social exchange theory, and interdependence theory for the recipient distinguishes these two methods of creating sup-
all suggest that altruism, generosity, and giving ultimately benefit port for the self by giving it to others. Support offered out of
the giver or the giver’s genes (Axelrod, 1984; Axelrod & Dion, compassion and concern for another should be accompanied by
1988; Buss & Kenrick, 1997; Rusbult & Van Lange, 2003; Thibaut positive other-directed emotions such as caring, affection, or love,
& Kelley, 1959). Norms of reciprocity, backed up with punishment whereas support given because of the expectation of reciprocity
for defection, encourage people to respond to cooperation with should not. The present studies examine the development of sup-
cooperation and to giving with giving in return. Thus, people can port in communal relationships, including friendships, family, and
elicit support from others by giving it first, obligating the support significant others, which are characterized by communal rather
recipient to reciprocate. Computer simulations and experimental than exchange norms.
evidence suggest that adding generosity to reciprocity, by respond-
ing somewhat more cooperatively than a partner’s previous re-
Goals and the Development of Support Over Time
sponse, can overcome misunderstanding and mistrust resulting
from “noisy” environments that interfere with direct reciprocation In the present investigations, we examine how goals shape the
(Kollock, 1993; Van Lange, Ouwerkerk, & Tazelaar, 2002). These development of support in communal relationships over time.
perspectives emphasize strategic giving with the (conscious or Goals reveal intentions, which may influence the effectiveness of
unconscious) expectation and intent to obtain something in return; support transactions (I. G. Sarason et al., 1990). Goals are also
supportive behavior is more a loan or an investment than a gift. malleable; consequently, goals could provide a useful point of
Relationships in which norms of reciprocity govern giving are intervention. In the present investigations, we examine two types
called exchange relationships (Clark, Mills, & Powell, 1986). of goals for the self in relationship to others that may shape the
People who are predisposed to cooperate in such relationships, development of social support: compassionate goals and self-
preferring to enhance the outcomes of a dyad, group, or collective image goals. We propose that support offered to others by people
while maintaining equality of outcomes instead of maximizing with compassionate goals builds a supportive environment char-
their own outcomes with little regard for others’ outcomes or acterized by feelings of closeness and connection, a cooperative
CREATING AND UNDERMINING SUPPORT 557
perspective, and decreased conflict. These positive effects of com- Self-Image Goals
passionate goals can be undermined, however, when the support
provider also has self-image goals focused on the construction of When people have self-image goals, they want to construct,
desired self-images. maintain, and defend desired public and private images of the self
to gain or obtain something for the self. In a great deal of research,
it has been demonstrated that people want to see themselves as
Compassionate Goals having desirable qualities (Bradley, 1978; Steele, 1988, p. 262;
Compassionate goals involve a focus on supporting others, not Taylor & Brown, 1988; Tesser, 1988). In social contexts, they also
to obtain something for the self, but out of consideration for the want others to recognize and acknowledge those qualities (Schlen-
well-being of others. When people have compassionate goals, they ker, 2003). Indeed, how people view themselves and how others
want to be a constructive force in their interactions with others, and view them are inextricably linked (Cooley, 1902/1956; Hardin &
they want to avoid harming others. Although people can differ, on Higgins, 1996; M. R. Leary & Downs, 1995; Shrauger & Schoen-
average, in how often they have compassionate goals in their eman, 1979; Tetlock & Manstead, 1985). Self-image goals involve
relationships with others, we hypothesize that people fluctuate self-presentation and impression management, not with the intent
from week to week, day to day, and possibly even moment to to deceive others but rather with the intent to convey an accurate,
moment in how compassionate their goals are. but idealized or glorified, conception of the self that the actor
We propose that people who are chronically high in compas- genuinely believes to be true (Baumeister, 1982; Greenwald &
sionate goals have personality characteristics, views of the self, Breckler, 1985; M. R. Leary, 1995; Schlenker, 1980). Constructing
relationship beliefs and styles, relationship experiences, and emo- desired self-images and getting others to recognize and acknowl-
tional states that distinguish them from people who are low in edge them help people accomplish interpersonal aims such as
compassionate goals. First, people with compassionate goals are winning friends, securing a job, or obtaining recognition (Schlen-
hypothesized to see people as interconnected and to feel caring and ker, 2003). We call this an ego-system perspective (Crocker, in
concerned for the well-being of others, regardless of identities such press).
as nationality or group membership. We refer to this as an eco- People who are chronically high in self-image goals have per-
system perspective, based on the biological notion of an ecosystem sonality characteristics, views of the self, relationship beliefs and
in which the well-being of one species is interdependent with the styles, and emotional states that distinguish them from people who
well-being of other species (Crocker, in press). This perspective is are low in self-image goals. First, we suggest that people with
reflected in two aspects of spiritual transcendence (Piedmont, self-image goals are primarily focused on their own needs and
1999): (a) universality, which refers to the feeling that all life is desires; they care about others only because others can give some
interconnected and to a the sense of shared responsibility of one desired social good to the self. Because they are trying to get
creature for another, and (b) connectedness, a sense of personal something from others, people with self-image goals should tend
responsibility to other people that extends across generations and to view their social interactions as zero sum in nature, with gains
within a community. Associated with this sense of interconnect- for one person coming at the expense of another; they should hold
edness, people with compassionate goals should tend to view individualistic beliefs about caregiving. Consequently, self-image
relationships with others as non–zero sum (Messick, 1967; Swin- goals may create conflict and loneliness. Because people with
gle, 1970); that is, they should assume that success for one person self-image goals want others to recognize and acknowledge their
does not detract from others; they should believe in mutual care- desirable qualities, they should attend to how the self is viewed by
giving and feel less psychological entitlement (i.e., not believe that others; that is, they should be high in public self-consciousness
they deserve more than others, Campbell, Bonacci, Shelton, Ex- (Fenigstein, 1987; Fenigstein et al., 1975). When people have
line, & Bushman, 2004). Their interpersonal goals should induce self-image goals, other people are potentially threatening—they
calm, positive, other-directed emotions such as love, connection, can provide or withhold something desired by the self. Conse-
and empathy. When people have compassionate goals, they should quently, we expect that self-image goals can foster feelings of
feel close to others. We also expected that people with compas- ambivalence, confusion, and fear.
sionate goals for others would be high in compassion toward
themselves; in other words, they are not caring toward others at the Implications of Compassionate Goals for Creating
expense of the self. Neff (2003) suggested that self-compassion Support
entails three basic components: (a) extending kindness and under-
standing toward oneself rather than harsh self-criticism and judg- When people have compassionate goals, we expect those people
ment; (b) seeing one’s suffering as part of the larger human to give more support to others, that others will feel supported, and
experience rather than as separating and isolating; and (c) mind- that others will want to give them support, not out of obligation but
fulness, which refers to keeping events and emotions in perspec- out of a sense of caring. In other words, people with compassionate
tive rather than blowing them out of proportion or exaggerating goals create a supportive interpersonal environment for others and
their significance. These qualities of self-compassion should be themselves. As a result, people who chronically have compassion-
revealed in private self-consciousness; that is, self-reflective ate goals may increase in trust, building confidence that others can
awareness of one’s internal states (Fenigstein, 1987; Fenigstein, be depended on and faith that others will be “responsive and caring
Scheier, & Buss, 1975). On the basis of these considerations, we despite the vicissitudes of an uncertain future” (Rempel, Holmes,
expect people who tend to have compassionate goals to be higher & Zanna, 1985, p. 97).
on the Big Five personality factors of Agreeableness and Extra- Intrapersonally, compassionate goals may directly increase per-
version than are people who are low in compassionate goals. ceived available social support through a process of projection. In
married couples, for example, perceived responsiveness to a support provider. Consequently, the support provider’s efforts to
spouse predicts perceptions of the spouse’s responsiveness to the be supportive, motivated by caring and compassion, will not
self more strongly than does the spouse’s self-reported responsive- induce caring and support from the recipient. As a result, upward
ness (LeMay et al., 2007). Thus, compassionate goals may lead to spirals of support and trust will not develop.
feelings of closeness, connection, and reduced conflict with others, In sum, we hypothesize that support providers’ compassionate
independent of others’ behavior. and self-image goals interact, such that self-image goals undercut
Interpersonally, compassionate goals may predict increased sup- the interpersonal benefits of compassionately motivated supportive
portive behaviors, which increase support recipients’ feelings of behaviors for support recipients and, consequently, the benefits of
being supported, increasing the recipient’s emotional connection to the behaviors for support providers.
the support provider and fostering supportive behavior, which in
turn increases the provider’s perceived available support, feelings Social Support and Adjustment to College
of closeness and connection, and reduced conflict. Because pro-
viders’ compassionate goals have positive consequences for their The processes we hypothesize should apply best to people who
own perceived available support, providers’ interpersonal trust and have recently experienced a disruption in social support. In their
desire to give support may also increase. In other words, we review of evidence that belongingness is a fundamental human
propose that providers’ perceived available social support and trust need, Baumeister and Leary (1995) suggested that once people
increase over time when they have compassionate goals. have reached a minimum number of social bonds that satisfy the
need to belong, additional bonds have little effect on well-being.
The Undermining Effect of Self-Image Goals Thus, the effect of goals on changes in social support and feelings
of closeness and loneliness should be most apparent among people
If this dynamic captured the complete story of goals and rela- experiencing a deficit of social support.
tionships, we would expect a general upward trend in social Students in their first semester at a college or a university
support over time among people with compassionate goals because provide an excellent population in which to test the effects of
no negative or countervailing effect blocks the upward spiral. compassionate goals on social support, particularly if the students
However, building on previous research on the self and on close do not live at home. The transition from high school to college is
relationships, we propose that self-image goals focused on self- challenging for a variety of reasons, including the increased diffi-
worth or self-image undermine these upward spirals of support and culty and competitiveness of academic work and the disruption of
trust (e.g., Collins & Feeney, 2000; Crocker & Park, 2004; B. C. social support networks involved in moving away from the nuclear
Feeney & Collins, 2001). In romantic relationships, egoistic mo- family for the first time. Along with meeting academic challenges,
tivations for caregiving negatively predict responsiveness of care- creating and maintaining friendships ranks among the most im-
giving and positively predicted controlling and compulsive care- portant tasks of the first semester of college.
giving (B. C. Feeney & Collins, 2001, 2003). Unmitigated In two studies, we examined links between compassionate and
communion, which involves sacrificing the self for the apparent self-image goals and change in social support and relationship
benefit of others, represents a type of self-image goal in which experiences among first-semester college freshmen. In Study 1, we
people want others to see them as giving but are actually motivated examined the intrapersonal effects of goals on change in perceived
by concerns about what their own self-worth is, whether others available support and interpersonal trust over the semester. In
will accept them, and how others evaluate them (Fritz & Helgeson, Study 2, we examined the interpersonal effects of goals on change
1998). in reports of supportive behaviors given and received over 3 weeks
Consistent with this research, we propose that supportive be- in a study of roommates who completed 21 daily reports of their
haviors are not always perceived as supportive because recipients goals for their roommate relationship.
may mistrust the intentions or goals of the support provider. As has
been seen, support is sometimes given out of concern for the Study 1
well-being of another and sometimes given strategically to obtain
something for the self. We propose that support recipients are In Study 1, we examined the effects of first-semester college
highly sensitive to the intentions of the support provider. Support students’ goals on changes in social support and trust over a
given to obtain something for the support provider creates an semester and weekly relationship experiences. We predicted that
obligation or perceived demand for the support recipient, and the compassionate goals would predict increases in perceived, avail-
needs of the recipient are secondary, if they are considered at all. able social support and trust, but only when participants are low in
Thus, support recipients may judge the goals of the provider. self-image goals. We assessed social support and trust early in the
When the provider signals a self-interested intention, supportive first semester and again at the end of the first semester. Between
behaviors may not feel supportive to the recipient. these two measurements, students completed 10 weekly measures
We propose that support providers’ self-image goals signal to of their goals for friendships, goal-related emotions, beliefs that
the support recipient that the support provider cares more about people should take care of each other or take care of themselves,
obtaining something for the self than about the well-being of the conflict, feelings of closeness and loneliness. This design allowed
recipient. Consequently, when support providers have compas- us to test the effects of chronic compassionate and self-image goals
sionate goals and give support but are also high in self-image averaged across the semester and the effects of week-to-week
goals, the support they provide will not be experienced as caring fluctuations in goals on relationship outcomes.
and supportive by recipients. Because recipients do not feel sup- We predicted that compassionate goals are associated with
ported, they should be less likely to be caring and responsive to the feeling clear and connected, feeling close to others, and believing
in mutual caregiving, but these effects should be undermined for and friends. These three types of social support may be affected
participants high in self-image goals. We predicted that self-image differently by goals in the first semester of college; whereas
goals are associated with feeling afraid and confused, feeling support from family may not depend on students’ friendship goals
lonely, individualistic caregiving beliefs, and interpersonal conflict in college, support from friends may be strongly related to stu-
and that these effects should be attenuated for participants high in dents’ friendship goals. Participants also completed a measure of
compassionate goals. The design of the study also allowed us to interpersonal trust at pretest and posttest. Additional measures
test whether these relationship experiences mediate the effects of assessed at pretest included spiritual transcendence (universality
the goals on changes in support and trust. As a secondary aim, in and connectedness subscales), zero-sum beliefs, narcissism, psy-
Study 1, we also examined associations of compassionate and chological entitlement, attachment anxiety and avoidance, Big
self-image goals with established personality, self, and relationship Five factors of personality, socially desirable responding, and
constructs. Consequently, at pretest and posttest, participants com- demographics (gender, race and ethnicity, age, parental income).
pleted several additional questionnaires. At posttest, participants completed measures of self-compassion,
public and private self-consciousness, and social anxiety. Unless
Method otherwise noted, participants rated their agreement with each item
on a scale ranging from 1 (strongly disagree) to 7 (strongly agree).
Participants Perceived social support availability. Perceived social support
Participants were 204 first-semester freshmen at a large Mid- availability was assessed in the pretest and the posttest with the
western university who volunteered for a study of goals and Multidimensional Survey of Perceived Social Support (Zimet,
adjustment to college within the first 3 weeks of the fall semester. Dahlem, & Zimet, 1988). Three subscales with four items each
Via advertisements in the campus newspaper, we offered partici- were used to assess perceived social support available from sig-
pants $60 for completing 12 surveys over the semester ($5 per nificant others, family, and friends. Internal reliabilities for signif-
survey) plus a $40 bonus for completing all 12 surveys. Five icant other, family, and friend subscales in this sample were .90,
participants who completed less than 3 surveys were dropped from .85, and .91, respectively, at pretest, and .94, .88, and .95, respec-
all analyses. Of the remaining 199 participants (122 or 61.3% tively, at posttest.
women; 77 or 38.7% men), 194 (97.5%) completed all 12 surveys. Interpersonal trust. Interpersonal trust was assessed in the
Of these participants, 141 (70.9%) reported their race as White or pretest and posttest with the interpersonal trust subscale from the
European American, 11 (5.5%) reported their race as Black or Eating Disorders Inventory (Garner, Olmstead, & Polivy, 1983).
African American, 37 (18.6%) reported their race as Asian or The subscale includes 7 items assessing trust of others, rated on a
Asian American, 3 (1.5%) reported their race as Latino(a), and 7 1 (never) to 6 (always) scale. The scale was internally reliable; ␣ ⫽
(3.5%) reported their race as other. The racial composition of the .87 at both pretest and posttest.
sample closely approximates the racial composition of the incoming Spiritual transcendence. Spiritual transcendence was assessed
freshman class. Participants ranged in age from 18 years to 21 years with two subscales from the Spiritual Transcendence Scale (Pied-
(M ⫽ 18.08 years). Modal parental income was $80,000 to $100,000 mont, 1999). The universality subscale includes five items assess-
per year, with 16 participants reporting parental income under ing a sense of interconnectedness among all life and all people.
$20,000 and 82 participants reporting parental income over $100,000. The connectedness subscale includes three items assessing the
sense of connection across time and place. The universality and
Procedure connectedness subscales had adequate internal reliability; ␣s ⫽ .87
and .67, respectively.
In groups of 4 to 12, participants attended a 1.5 hr session to
In zero-sum situations, one person’s good outcomes come at the
learn about the study, give their informed consent, complete the
expense of another person (e.g., Messick, 1967). In the present
pretest survey, and receive instructions for completing the remain-
study, we created a six-item measure of zero-sum beliefs. The
ing 11 surveys. Participants completed the next 10 surveys on the
items included, “One person’s success depends on another per-
Internet, 1 survey per week for 10 weeks. The weekly and posttest
son’s failure,” “In order to succeed in this world, it is sometimes
surveys were administered with UM Lessons (Version 2K10.2)
necessary to step on others along the way,” “My successes don’t
software, an online survey tool. To protect their identities, partic-
mean much if most other people succeed at the same task,” “An
ipants entered randomly assigned codes upon accessing the survey.
accomplishment is only really meaningful if it is rare,” “To give to
The weekly surveys took about 30 min to complete. Participants
others usually requires a sacrifice on the part of the giver,” and “I
received an e-mail reminder with a link to the survey each week.
believe that people are basically self-interested.” Exploratory fac-
Participants who did not complete the survey on the designated
tor analyses indicated that the items loaded on a single factor, and
day received a follow-up reminder. To retain as many participants
the scale was reliable (␣ ⫽ .77).
as possible in the study, participants who missed a week could
Narcissism. Narcissism was assessed with the Narcissistic
complete the survey the following week, extending the duration of
Personality Inventory, a 40-item forced-choice measure (Raskin &
the study by a week, not to extend beyond the fall semester. Once
Terry, 1988). The scale was internally reliable (␣ ⫽ .83).
participants had completed all 10 weekly surveys, they completed the
Psychological entitlement. Psychological entitlement was as-
posttest survey and were debriefed and paid for their participation.
sessed with the Psychological Entitlement Scale (Campbell et al.,
2004). The nine items assess the feeling of deserving more than others
Pretest and Posttest Measures
deserve. The entitlement scale had high internal reliability (␣ ⫽ .89).
At pretest and posttest, participants completed measures of The Self-Consciousness Scale. The Self-Consciousness Scale
perceived available social support from significant others, family, (Fenigstein et al., 1975) has three subscales: a 10-item measure of
private self-consciousness, which assesses self-reflection; a 7-item did you want to or try to,” and items were rated on a scale ranging
measure of public self-consciousness, which assess the tendency to from 1 (not at all) to 5 (always). Seven items assessed compas-
focus on how others view the self; and a 6-item measure of social sionate goals, namely, “be supportive of others,” “have compas-
anxiety. The scale has demonstrated reliability and validity (Fenig- sion for others’ mistakes and weaknesses,” “avoid doing anything
stein, 1987; Fenigstein et al., 1975). In the present study, all three that would be harmful to others,” “make a positive difference in
subscales had good reliability: public, ␣ ⫽ .85; private, ␣ ⫽ .74; someone else’s life,” “be constructive in your comments to oth-
and anxiety, ␣ ⫽ .82. ers,” “avoid being selfish or self-centered,” and “avoid doing
Self-compassion. Self-compassion was assessed with a mea- things that aren’t helpful to me or others.” Six items assessed
sure developed by Neff (2003). The Self-Compassion Scale has six self-image goals, namely, “get others to recognize or acknowledge
subscales assessing self-kindness (e.g., “When I’m going through your positive qualities,” “convince others that you are right,”
a very hard time, I give myself the caring and tenderness I need”), “avoid showing your weaknesses,” “avoid the possibility of being
self-judgment (e.g., “I’m disapproving and judgmental about my wrong,” “avoid being rejected by others,” and “avoid taking risks
own flaws and inadequacies”), common humanity (e.g., “When or making mistakes.” Exploratory factor analyses with principal
things are going badly for me, I see the difficulties as part of life factors extraction and oblimin rotation, which allows factors to be
that everyone goes through”), isolation (e.g., “When I’m really correlated (Russell, 2002), indicated that although the factor load-
struggling, I tend to feel like other people must be having an easier ings varied somewhat from week to week, when averaged across
time of it”), mindfulness (e.g., “When something upsets me I try to the 10 weeks, the items loaded cleanly (self-image: all loadings ⱖ
keep my emotions in balance”), and overidentification (e.g., .52; compassionate: all loadings ⱖ .66). With one exception (“get
“When I’m feeling down I tend to obsess and fixate on everything others to recognize or acknowledge your positive qualities,” which
that’s wrong”). Each of the subscales had good internal reliability: loaded .23 on the compassionate factor), cross-loadings did not
the total scale (after reverse scoring self-judgment, isolation, and exceed .20, and the average loading on the intended factor always
overidentification) and the self-kindness, self-judgment, and com- exceeded the average loading on the second factor by a ratio of 2:1
mon humanity subscales each had ␣ ⫽ .83; the isolation (␣ ⫽ .81), or greater. Both scales had high internal consistency each week of
overidentification (␣ ⫽ .82), and mindfulness (␣ ⫽ .77) subscales the study (self-image goals: .77 ⬍ ␣ ⬍ .89, M␣ ⫽ .83; compas-
also had good reliability. sionate goals: .83 ⬍ ␣ ⬍ .93, M␣ ⫽ .90).1
Attachment anxiety and avoidance. Attachment anxiety and Goal-related affect. Each week, participants answered an
avoidance were assessed at pretest with abbreviated versions of the open-ended question about their friendship goals: “Thinking about
measure developed by Brennan, Clark, and Shaver (1998) to assess your friendships (meeting people, partying, making friends, being
attachment styles in the context of romantic relationships. The a friend, etc.) what was your main goal in the past week?”
original scale includes 18 items assessing each dimension; we Participants then rated the extent to which having this goal made
included the 9 highest loading items for each subscale. The result- them feel critical (of self or others), competitive, clear, loving,
ing measures had high internal consistency (␣s ⫽ .88 and .93 for confused, peaceful, connected to others, empathic, isolated, en-
attachment anxiety and avoidance, respectively). gaged, present, ambivalent/conflicted, pressured, distracted, coop-
Big Five. The Big Five factors of personality were assessed erative, and fearful, on a scale ranging from 1 (not at all) to 5
with a brief measure designed for use when time constraints (extremely). Factor analysis of these items with principal factors
prevent the use of longer measures (Gosling, Rentfrow, & Swann, extraction and oblimin rotation yielded two factors, which we
2003). The brief measure includes two items assessing each Big called clear and connected (peaceful, connected to others, coop-
Five factor. Although the subscales are less reliable than longer
erative, loving, clear, present, empathic, and engaged; ␣ ⫽ .91),
measures of Big Five factors, they show adequate convergent and
and afraid and confused (fearful, ambivalent/conflicted, pressured,
discriminant validity, test–retest reliability, patterns of external
distracted, confused, critical, isolated, and competitive; ␣ ⫽ .90).
correlates, and convergence between self- and observer-ratings
Closeness, loneliness, conflict, and beliefs about mutual and
(Gosling et al., 2003). In the present study, reliabilities were
individualistic caregiving. Closeness, loneliness, conflict, and
obtained for the following: extraversion (␣ ⫽ .68), agreeableness
beliefs about mutual and individualistic caregiving were each
(␣ ⫽ .46), conscientiousness (␣ ⫽ .58), emotional stability (␣ ⫽
assessed with single items created for this study: “In the past week,
.56), and openness to experience (␣ ⫽ .28).
how often did you ‘feel close to others,’ ‘feel lonely,’ ‘have
Social desirability. Social desirability was measured at pretest with
conflicts with people,’ ‘feel it was important that people look out
the Marlowe–Crowne Social Desirability Scale (Crowne & Marlowe,
for one another,’ and ‘feel it was important to look out for yourself,
1964), a 33-item forced-choice scale designed to assess the extent to
even at the expense of others?’” The items were rated on scales
which people describe themselves in favorable terms (␣ ⫽ .77).
ranging from 1 (never) to 5 (always).
Weekly Measures 1
Both scales included approach and avoidance items to avoid confound-
ing compassionate and self-image goals with approach and avoidance
Each of the 10 weekly assessments included measures of com-
goals. In factor analyses of these items, two factors emerge, with self-
passionate and self-image goals, goal-related emotion, hostility,
image approach and avoidance items loading on one factor and compas-
closeness, loneliness, conflict, and zero-sum and non–zero-sum sionate approach and avoidance items loading on the other factor. As
orientations. expected, when we created separate approach and avoidance subscales for
Compassionate and self-image goals. Compassionate and compassionate and self-image goals, approach and avoidance self-image
self-image goals were assessed with 13 items. All items began with goals correlated highly (r ⫽ .64, p ⬍ .001), and approach and avoidance
the phrase, “In the past week, in the area of friendships, how much compassionate goals correlated highly (r ⫽ .76, p ⬍ .001).
Results tions in goals and their interaction predict within-person changes

in clear and connected affect, afraid and confused affect, closeness
Overview of Analyses and loneliness, conflict, individualistic caring, and belief in mutual
We computed chronic compassionate and self-image goals by caring.
averaging across the 10 weekly reports. Data analyses proceeded
in four phases. In Phase 1, we examined the convergent and Correlates of Compassionate and Self-Image Goals
divergent validity of the goals with the beliefs, self-relevant vari- We first examined associations between compassionate and
ables, relationship style variables, and Big Five personality factors self-image goals and the individual difference variables assessed
assessed at pretest or posttest. In Phase 2, we examined whether as possible covariates. Because compassionate goals correlated
goals predict change in perceived available social support and with self-image goals (r ⫽ .53, p ⬍ .001), the zero-order correla-
interpersonal trust over the semester. In a series of analyses, we tions are difficult to interpret because they could be spurious.
then entered correlates of goals from Phase 1 to see whether any of Consequently, we also entered the goals simultaneously in regres-
them accounted for the effects in Phase 2. In Phase 3, we examined sion analyses to assess the unique effect of each goal while
whether chronic goals or their interaction predicted average levels controlling for the other. Social desirability correlated negatively
of weekly clear and connected affect, afraid and confused affect, with self-image goals (r ⫽ ⫺.15, p ⬍ .05) and positively with
closeness, loneliness, conflict, and beliefs in mutual and individ- compassionate goals (r ⫽ .30, p ⬍ .01). Men (M ⫽ 2.93) and
ualistic caregiving; we also examined whether average weekly women (M ⫽ 3.04) did not differ significantly on self-image goals,
outcomes on these variables mediate the effects of goals on change but did differ on compassionate goals (Mmen ⫽ 3.14, Mwomen ⫽
in social support and trust. In Phase 4, using hierarchical linear 3.42, t ⫽ ⫺2.87, p ⬍ .01). Consequently, in all regression analyses
modeling (HLM), we examined whether within-person fluctua- we controlled for gender and social desirability.
Table 1
Zero-Order Correlations and Regression Coefficients Predicting Individual Differences From
Compassionate and Self-Image Goals
Compassionate Self-image Compassionate Self-image

Variable goals r goals r goals ␤ goals ␤
Spiritual transcendence
Universality .48*** .33*** .38*** .13
Connectedness .38*** .21** .36*** .00
Zero sum beliefs ⫺.14 .19** ⫺.18** .26**
Self-compassion
Total score .12 ⫺.22** .38*** ⫺.39***
Low isolation ⫺.09 ⫺.31*** .11 ⫺.33***
Low self-judgment ⫺.13 ⫺.39*** .13 ⫺.42***
Mindfulness .35*** .02 .54*** ⫺.26**
Low over-identified ⫺.11 ⫺.33*** .11 ⫺.34***
Self-kindness .24** ⫺.02 .42*** ⫺.22*
Common humanity .35*** .13 .45*** ⫺.15
Self-consciousness
Public .27*** .59*** ⫺.06 .60***
Private .28*** .10 .40*** ⫺.14
Social anxiety .04 .19** ⫺.09 .22*
Attachment style
Anxious .05 .30*** .03 .23***
Avoidant ⫺.06 .15* ⫺.22* .25***
Narcissism (NPI)
Total ⫺.13 .05 ⫺.15 .13
Arrogance ⫺.13 ⫺.05 ⫺.18 .06
Entitlement ⫺.23** .05 ⫺.14 .09
Leadership ⫺.04 .04 ⫺.05 .06
Self-admiration ⫺.05 .10 ⫺.11 .15
Psychological ⫺.22** .07 ⫺.28** .20*
entitlement
Big Five
Agreeableness .40*** .09 .30*** ⫺.02
Extraversion .30*** .19** .29** .02
Openness .17* ⫺.00 .15 ⫺.06
Conscientiousness .06 .08 ⫺.06 .11
Emotional Stability .10 ⫺.11 .10 ⫺.10
Note. The table shows betas with both goals entered simultaneously, with gender and social desirability
controlled for. NPI ⫽ Narcissistic Personality Inventory.
*
p ⬍ .05. ** p ⬍ .01. *** p ⬍ .001.
Table 1 presents the zero-order correlations between chronic
goals were calculated by averaging across the 10 weekly reports. Social support was measured on a scale ranging from 1 (strongly disagree) to 7 (strongly agree). Interpersonal trust was measured
Note. N ⫽ 199 at pretest (below diagonal); N ⫽ 194 at posttest (above diagonal). Pretest–posttest correlations are reported on the diagonal. Closeness, loneliness, and compassionate and self-image
on a scale ranging from 0 (never) to 4 (always). Chronic closeness, loneliness, affect, conflict, belief in mutual caring, individualistic caring, self-image goals, and compassionate goals were measured
1.45
1.25
1.21
1.01
SD
Posttest
compassionate and self-image goals and individual difference vari-
ables assessed as possible covariates, as well as regression coef-
5.46
5.40
5.58
2.89
M
ficients when both goals are entered as predictors. Controlling for
self-image goals, compassionate goals were associated with high
1.28
1.26
1.02
0.92
0.71
0.82
0.64
0.68
0.63
0.76
0.79
0.67
0.63
SD
Pretest
spiritual transcendence (both universality and connection), low
zero-sum beliefs, high self-compassion (especially mindfulness,
5.71
5.52
5.82
2.89
3.40
2.65
2.22
2.86
2.19
3.66
2.41
3.31
3.00
M
self-kindness, and common humanity), high private self-
consciousness, low avoidant attachment style, low psychological
.28***
13
—
entitlement, and high agreeableness and extraversion. Self-image
.07
.06
.07
goals, on the other hand, are uniquely associated with higher
.35***
.53***
.41***
.54***
zero-sum beliefs, lower self-compassion (significant for all sub-
.23**
12
—
scales except common humanity), higher public self-consciousness
and social anxiety, increased attachment insecurity (both anxiety
⫺.29***
⫺.19**
⫺.20**
and avoidance), and greater psychological entitlement.
.15*
11
—
⫺.11
⫺.10
Table 2 shows the means, standard deviations, and intercorre-
lations among the main variables; where relevant, correlations
.30***
.26***
.51***
.43***
⫺.42***
.75***
.23**
among pretest variables are below the diagonal and correlations
10
—
among posttest variables are above the diagonal, with test–retest
reliabilities on the diagonal.
.31***
.26***
⫺.18*
⫺.15*
—
⫺.13
⫺.08
⫺.13
⫺.03
9
Zero-Order Correlations, Means, and Standard Deviations of Main Variables in Regression Analyses in Study 1
Change in Social Support and Trust From Pretest to
Posttest
.32***
.43***
.39***
.52***
.69***
.44***
.21**
—
.03
⫺.07
8
We examined the associations between goals and change in
social support and trust from pretest to posttest. We regressed the
posttest outcomes on pretest measure, compassionate goals, and
.38***
.48***
.53***
.20**
.21**
—
⫺.01
⫺.01
.09
⫺.08
.07
self-image goals. To test whether self-image goals interacted with
7
compassionate goals to predict posttest residual outcomes, we

grand mean centered goals to reduce multicolinearity (Aiken &
⫺.28***
.46***
.37***
.39***
⫺.15*
.17*
—
⫺.13
⫺.07
.04
.04
.09
West, 1991), then entered the Compassionate ⫻ Self-Image goal
6
product term as a second block in a hierarchical regression anal-

ysis. Table 3 shows the standardized regression coefficients for the
.46***
.33***
.69***
.68***
.59***
.56***
.63***
.28***
⫺.24**
associations between goals and relationship variables at pretest,
—
⫺.22
.05
⫺.04
5
posttest, and posttest controlling for pretest.

Main effects of goals on change in social support. When we
.47***
.31***
.67***
.76***
.55***
⫺.26***
.36***
.34***
.34***
regressed posttest social support on goals, controlling for pretest ⫺.18*
⫺.08
.13
.02
4
support to examine change, compassionate goals predicted in-

creased social support from pretest to posttest. As Table 3 shows,
.61***
.39***
.56***
.47***
.44***
.29***
.29***
.33***
.19**
this pattern of results held for all three measures of social support,
⫺.15*
⫺.08
.06
⫺.09
3
with the largest increase observed for social support from friends.
Self-image goals predicted decreased social support from pretest to
.43***
.72***
.31***
.24***
.33***
.24***
posttest, but this effect reached statistical significance only for

.22**
.17*
.14*
⫺.09
.08
⫺.08
⫺.12
2
social support from family.

Interaction effects. We next examined whether goals interact
on a scale ranging from 1 (never) to 5 (always).
.53***
.34***
.51***
.43***
.41***
.27***
.24***
to predict change in social support from pretest to posttest. In

.19**
⫺.10
.04
⫺.10
⫺.13
.06
1
hierarchical regression analyses, entering the interaction term sig-

p ⬍ .05. ** p ⬍ .01. *** p ⬍ .001.
nificantly increased the variance explained in posttest social sup-

port from friends (⌬R2 ⫽ .045, p ⬍ .001) and significant others
Individualistic caregiving belief
(⌬R2 ⫽ .021, p ⬍ .05) but not family (⌬R2 ⫽ .003, ns). Following
the recommendations of Aiken and West (1991), we computed the
Significant other support
Mutual caregiving belief

Afraid/confused affect
Clear/connected affect
simple slopes for the effect of compassionate goals at high and low
Compassionate goals
levels of self-image goals (1 SD above and below the mean).

Variable
Interpersonal trust
Self-image goals
Family support
Figure 1 shows the predicted values for compassionate goals

Friend support
predicting residual posttest social support from friends and from

Loneliness
Closeness
significant others at higher and lower levels of self-image goals.

Conflict
For participants low in self-image goals, compassionate goals

Table 2
predicted increased social support from friends (␤ ⫽ .41, p ⬍ .001)

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
and significant others (␤ ⫽ .26, p ⬍ .001). For participants high in

*
Table 3
Standardized Regression Coefficients for the Associations Among Mean Weekly Compassionate and Compassionate Goals and
Relationship Variables at Pretest and Posttest and Residual Change From Pretest to Posttest in Study 1
Significant other support Family support Friend support Interpersonal trust
Goal Pretest Posttest Change Pretest Posttest Change Pretest Posttest Change Pretest Posttest Change
** *** *** ** *** ** ** *** *** *** ***
Compassionate .27 .42 .31 .25 .33 .18 .30 .59 .47 .46 .55 .26***
Self-image ⫺.11 ⫺.18* ⫺.13 ⫺.01 ⫺.15 ⫺.16* .02 ⫺.07 ⫺.11 ⫺.22** ⫺.23* ⫺.10
Note. N ⫽ 199 at pretest; N ⫽ 194 at posttest. Goals were entered simultaneously; all regression analyses control for gender and social desirability.
*
p ⬍ .05. ** p ⬍ .01. *** p ⬍ .001.
self-image goals, compassionate goals did not significantly predict Chronic Goals and Mean Weekly Outcomes
increased social support from friends (␤ ⫽ .09, ns) or significant
As expected, averaging across the 10 weekly reports, students who
others (␤ ⫽ .05, ns). Figures 1A and 1B depict the predicted values
reported higher compassionate goals reported greater clear and con-
for these interactions (Aiken & West, 1991).
nected affect, closeness, and mutual caregiving beliefs, whereas stu-
Controlling for related constructs. To determine whether re-
dents who reported higher self-image goals reported greater afraid and
sidual change in perceived available social support associated with
confused affect, loneliness, individualistic caregiving beliefs, and
goals was due to the association of goals with other constructs, we
conflict. Compassionate goals positively predicted feeling clear and
conducted a series of regression analyses predicting posttest sup-
connected (␤ ⫽ .61, p ⬍ .001), but self-image goals did not (␤ ⫽ .11,
port from compassionate and self-image goals, controlling for
ns). Entering the interaction in a second block of the hierarchical
gender, social desirability, and covariates. In eight separate anal-
regression equation significantly increased the variance explained
yses for each support outcome, we controlled for each set of
(⌬R2 ⫽ .021, p ⬍ .01). Compassionate goals were more strongly
covariates in Table 1. In each of these analyses, the effect of
associated with greater clear and connected feelings for students low
compassionate goals on residual change in perceived available
in self-image goals (␤ ⫽ .63, p ⬍ .001) than for students high in
social support remained highly significant, with one exception:
self-image goals (␤ ⫽ .38, p ⬍ .001).
self-compassion predicted residual change in family support and
Self-image goals positively predicted feeling afraid and con-
reduced the effects of compassionate goals from .18 to .13.2 The
fused (␤ ⫽ .55 p ⬍ .001), but compassionate goals did not (␤ ⫽
interaction effect remained significant for friend support ( ps ⬍
⫺.07, ns). Entering the Self-Image ⫻ Compassionate interaction in
.05), and for significant other support ( ps ⬍ .05). For family
a second block of the hierarchical regression equation increased
support, the interaction effect remained nonsignificant regardless
the variance explained (⌬R2 ⫽ .016, p ⬍ .05). Self-image goals
of which covariates were included.
were more strongly associated with afraid and confused feelings
Interpersonal trust. Regression analyses predicting posttest
for students low in compassionate goals (␤ ⫽ .62, p ⬍ .001) than
trust controlling for pretest trust indicated that compassionate
for students high in compassionate goals (␤ ⫽ .42 p ⬍ .001).
goals predict increased interpersonal trust from pretest to posttest
Closeness related positively with compassionate goals (␤ ⫽ .69,
(see Table 3). Entering the interaction into the regression equation
p ⬍ .001) but not self-image goals (␤ ⫽ ⫺.11, ns). Entering the
in a second block of the hierarchical regression significantly in-
Self-Image ⫻ Compassionate interaction in a second block of a
creased the variance explained (⌬R2 ⫽ .019), p ⬍ .01. For partic-
ipants low in self-image goals, compassionate goals significantly 2
predicted increases in trust from pretest to posttest (␤ ⫽ .41, p ⬍ For friendship support, with covariates entered, the effect of compas-
sionate goals ranged from .36 ⬎ ␤s ⬎ .49, ps ⬍ .001. The effect of
.001); for participants high in self-image goals, compassionate
self-image goals remained nonsignificant. Several variables predicted re-
goals did not significantly predict increases in trust (␤ ⫽ .08, ns). sidual change in friend support, controlling for gender, social desirability,
Figure 2 depicts predicted values of posttest trust for participants and goals: self-compassion (␤ ⫽ .16, p ⬍ .01), public self-consciousness
high (1 SD above the mean) and low (1 SD below the mean) in (␤ ⫽ .16, p ⬍ .03), social anxiety (␤ ⫽ ⫺.20, p ⬍ .001), and extraversion
self-image goals. Self-image goals did not predict change in trust. (␤ ⫽ .21, p ⬍ .001). For significant other support, with the covariates
The main effect of compassionate goals ( ps ⬍ .005) and interac- entered, the effect of compassionate goals remained significant (.25 ⬎
tion ( ps ⬍ .05) remained significant when we entered the covari- ␤s ⬎ .31, ps ⬍ .01); the effect of self-image goals remained nonsignificant.
ates in Table 1 in separate regression analyses.3 In sum, students Only private self-consciousness explained variance in residual change in
with high compassionate goals and low self-image goals showed significant other support beyond that accounted for by gender, social
increases in perceived social support from friends and significant desirability, and goals (␤ ⫽ .13, p ⬍ .05). For family support, self-
compassion significantly predicted residual change (␤ ⫽ .14, p ⬍ .02), and
others and increases in interpersonal trust in their first semester of
controlling for self-compassion reduced the effect of compassionate goals
college. These results are consistent with our hypothesis that to ␤ ⫽ .13, p ⬍ .06, and self-image goals to nonsignificance (␤ ⫽ ⫺.11,
compassionate and self-image goals interact to affect change in p ⬍ .11). In all other analyses, the effect of compassionate goals (.15 ⬎
social support and trust over the first semester of college. Students ␤s ⬎ .19, ps ⬍ .05) and self-image goals (.15 ⬎ ␤s ⬎ .17, ps ⬍ .05)
with high compassionate goals and low self-image goals show the remained significant.
highest levels of social support and trust at the end of their first 3
Only total self-compassion (␤ ⫽ .20, p ⬍ .001) and social anxiety (␤ ⫽
semester of college, controlling for pretest levels of these vari- ⫺.10, p ⬍ .05) explained unique variance in residual change in trust
ables. beyond that explained by goals, gender, and social desirability.
Low Self-Image Goals

High Self-Image Goals
A B
6.5 6.5
Residual Posttest Significant Other Support

Residual Posttest Friend Support
6 6
5.5 5.5
5 5
Low High Low High
Compassionate Goals Compassionate Goals
Figure 1. Compassionate goals predicting residual posttest social support (scores) from (A) friends and (B)
social support from significant others, at higher and lower levels of self-image goals in Study 1. Means are
plotted at 1 SD above and 1 SD below the mean for self-image goals and compassionate goals. In all analyses,
we control for gender, social desirability, and the corresponding pretest variable.
hierarchical regression equation significantly increased the vari- were stronger when participants were low in self-image goals.
ance explained (⌬R2 ⫽ .039, p ⬍ .001). Compassionate goals were Self-image goals were strongly associated with average afraid and
associated with closeness at both high and low levels of self-image confused feelings and conflict; these effects were stronger when
goals, but the effect of compassionate goals on closeness was participants were low in compassionate goals. In addition, com-
stronger for students who were low (␤ ⫽ .71, p ⬍ .001) rather than passionate goals and self-image goals predicted conflict and be-
high (␤ ⫽ .38, p ⬍ .001) in self-image goals. liefs about mutual caring and individualistic caregiving, although
Loneliness related positively with self-image goals (␤ ⫽ .40, the goals did not interact to predict these beliefs.4
p ⬍ .001) but not compassionate goals (␤ ⫽ ⫺.13, ns). Entering
the Self-Image ⫻ Compassionate interaction in a second block of Mediation Analyses
the hierarchical regression equation significantly increased the
variance explained (⌬R2 ⫽ .032, p ⬍ .01). Self-image goals were To see whether any of the average weekly measures could
more strongly associated with loneliness for students low in com- explain the effects of goals on change in social support and trust,
passionate goals (␤ ⫽ .54, p ⬍ .001) than for students high in we conducted four additional regression analyses for each outcome
compassionate goals (␤ ⫽ .29, p ⬍ .01). controlling for (a) clear and connected feelings and afraid and
Average weekly conflict related positively with self-image goals confused feelings, (b) closeness and loneliness feelings, (c) con-
(␤ ⫽ .27, p ⬍ .01), but not compassionate goals (␤ ⫽ ⫺.10, ns). flict, and (d) beliefs that people should look out for each other and
Entering the Self-Image ⫻ Compassionate interaction in a second look out for themselves.
block did not significantly increase the variance explained (⌬R2 ⫽ Friend support. Closeness significantly predicted residual
.012, ns). When we entered the covariates in Table 1 in separate change in friend support (␤ ⫽ .46, p ⬍ .001), and the effect of
analyses, the main effect of self-image goals remained significant; compassionate goals was reduced significantly, from ␤ ⫽ .46, p ⬍
the main effect of compassionate goals remained nonsignificant.
4
Compassionate goals positively predicted belief in mutual care- All significant main effects of compassionate and self-image goals and
giving (␤ ⫽ .84, p ⬍ .001); self-image goals negatively predicted significant interaction effects in these analyses remained significant when we
this belief (␤ ⫽ ⫺.21, p ⬍ .001). The Self-Image ⫻ Compassion- controlled for the covariates in Table 1 (eight regression analyses for each
ate interaction did not significantly increase the variance explained average weekly outcome), with the following exceptions For clear and con-
nected, with public and private self-consciousness and social anxiety con-
when entered on the next block (⌬R2 ⫽ .00). Self-image goals
trolled, the interaction effect was reduced to ␤ ⫽ ⫺.12, p ⫽ .053. For
positively predicted individualistic caregiving belief (␤ ⫽ .34, p ⬍
closeness, The main effect of self-image goals was reduced to nonsignificance
.001); compassionate goals negatively predicted this belief (␤ ⫽ when we included covariates in the analyses (⫺.13 ⱖ ␤s ⬎ .09, ps ⬎ .05). For
⫺.38, p ⬍ .001). Entering the interaction term did not significantly conflict, the Self-Image ⫻ Compassionate goals interaction was reduced to
increase the variance explained (⌬R2 ⫽ .003, ns). marginal significance when we controlled for self-compassion, self-
In sum, compassionate goals were strongly associated with consciousness, narcissism, or entitlement. It was reduced to nonsignificance
average close and connected feelings and closeness; these effects when we controlled for agreeableness, extraversion, and openness.
Low Self-Image Goals ⫺.13, p ⬍ .05), and the Self-Image ⫻ Compassionate goals
High Self-Image Goals interaction remained nonsignificant ( p ⬎ .28).
5
Trust. Only closeness predicted residual change in trust (␤ ⫽
.40, p ⬍ .001), and with closeness controlled, the effect of com-
passionate goals became nonsignificant, indicating full mediation
(␤ ⫽ .05, p ⬎ .40; Sobel’s Z ⫽ 5.10, p ⬍ .001). The interaction
between self-image goals and compassionate goals remained sig-
Residual Posttest Trust
4.5 nificant, p ⬍ .05.

In sum, mediation analyses indicated that either average weekly
closeness or average weekly mutual caring beliefs, or both to-
gether, mediated the main effect of compassionate goals on change
in all three types of support and trust. The alternative model, in
4
which closeness affects support or trust via its effect on compas-
sionate goals, was not supported. None of the average weekly
variables explained the interactive effects of compassionate and
self-image goals, however.
3.5
Low High
Compassionate Goals
Weekly Changes in Compassionate and Self-Image Goals
The preceding analyses do not address whether increases in
Figure 2. Compassionate goals predicting residual posttest trust (scores)
at higher and lower levels of self-image goals in Study 1. Means are plotted
compassionate or self-image goals from week to week predict
at 1 SD above and 1 SD below the mean for self-image goals and increased relationship outcomes in those weeks. In HLM analyses,
compassionate goals. we investigated the within-person effects of goals. In HLM, mul-
tiple levels of nested data are examined simultaneously. In our
data, weekly reports are the lower level (or Level 1) units, which
are nested within persons (the Level 2 units). Multilevel modeling
.001 to ␤ ⫽ .17, p ⬍ .05, indicating partial mediation (Sobel’s Z ⫽ is particularly suited for analyzing these data because it can effec-
5.58, p ⬍ .001). In a separate analysis, mutual caring beliefs also tively manage missing data on the repeated measure and can adjust
predicted residual change in friend support (␤ ⫽ .26, p ⬍ .01), and for any bias in standard errors and statistical tests resulting from
the effect of compassionate goals was reduced (␤ ⫽ .25, p ⬍ .05), the nonindependence of observations (Kenny, Korchmaros, &
suggesting partial mediation (Sobel’s Z ⫽ 3.32, p ⬍ .001). Feel- Bolger, 2003; Krull & MacKinnon, 2001). In multilevel modeling,
ings (clear and connected or afraid and confused), conflict, and the within-person effects can vary randomly across persons, and
loneliness did not significantly predict change in friend support. person-level (Level 2) variables may explain variance in the
When we entered both closeness and belief in mutual caregiving in within-person effects.
the equation together, each predicted unique variance in change in All HLM analyses were conducted with restricted maximum
friend support (␤closeness ⫽ .44, p ⬍ .001; ␤mutual caring ⫽ .19, p ⬍ likelihood models. Level 1 variables were group mean centered, so
.05), and the effect of compassionate goals was reduced to non- effects reflect changes in weekly variables compared with partic-
significance (␤ ⫽ .03, p ⬎ .7). The Self-Image ⫻ Compassionate ipants’ own average levels, as a function of changes in their goals
goals interaction remained significant (⌬R2 ⫽ .02, p ⬍ .01). These compared with their own average levels. We entered compassion-
findings indicate that average weekly closeness and mutual caring ate and self-image goals simultaneously and then entered the
beliefs together completely mediate the main effect of compas- interaction.
sionate goals on change in perceived available support from Within- and between-persons variance. Both goals varied
friends. The alternative model, in which closeness affects change more between persons than within persons. For compassionate
in friend support via compassionate goals is not plausible because goals, 69% of the total variance was between persons, and 31%
the path from compassionate goals to change in friend support was occurred within persons. For self-image goals, 67% of the total
not significant (␤ ⫽ .03, p ⬎ .7). variance was between persons, and 32% occurred within persons.
Significant other support. Closeness predicted residual change Goal-related affect. Weekly increases in compassionate goals
in significant other support (␤ ⫽ .17, p ⬍ .05), and the effect of predicted weekly increases in feeling clear and connected (B ⫽
compassionate goals was reduced from ␤ ⫽ .30, p ⬍ .001 to ␤ ⫽ 0.45, p ⬍ .001), but weekly increases in self-image goals did not.
.18, p ⬍ .07 (Sobel’s Z ⫽ 2.30, p ⬍ .05). None of the other average The Self-Image ⫻ Compassionate interaction significantly pre-
weekly measures predicted change in significant other support. dicted changes in feeling clear and connected (B ⫽ ⫺0.09, p ⬍
The Self-Image ⫻ Compassionate goals interaction remained sig- .001). Computing the simple slopes revealed that weekly increases
nificant ( ps ⬍ .02). in compassionate goals predicted weekly increases in feeling clear
Family support. Only the belief that people should take care of and connected more strongly on weeks when participants were low
each other predicted change in family support with goals in the in self-image goals (B ⫽ 0.50, p ⬍ .001) than on weeks when they
equation (␤ ⫽ .17, p ⬍ .05), and with this variable controlled, the were high in self-image goals (B ⫽ 0.36, p ⬍ .001).
effect of compassionate goals became nonsignificant (␤ ⫽ .03, Increases in self-image goals predicted weekly increases in
p ⬎ .73), indicating full mediation (Sobel’s Z ⫽ 2.22, p ⬍ .05). feeling afraid and confused (B ⫽ ⫺0.24, p ⬍ .001), and increases
The main effect of self-image goals remained significant (␤ ⫽ in compassionate goals predicted decreases in feeling afraid and
confused (B ⫽ ⫺0.10, p ⬍ .01). The Self-Image ⫻ Compassionate vidual difference covariates, from Big Five factors of personality
interaction significantly predicted changes in feeling afraid and to spiritual transcendence, compassionate goals still predicted in-
confused (B ⫽ ⫺0.09, p ⬍ .001). Computing the simple slopes creases over the semester. Only family support, which showed the
revealed that weekly increases in compassionate goals predicted smallest change, was reduced to nonsignificance with the addition
weekly decreases in feeling afraid and confused on weeks when of a covariate: self-compassion.
participants were high in self-image goals (B ⫽ ⫺0.19, p ⬍ .001), Third, students with high compassionate goals over the semester
but not on weeks when they were low in self-image goals (B ⫽ also reported that their friendship goal made them feel clear and
⫺0.06, ns). connected, more closeness, more belief in mutual caring, and less
Weekly closeness and loneliness. Weekly increases in self- conflict averaged over 10 weeks. Students with high self-image
image goals predicted increases in loneliness (B ⫽ 0.24, p ⬍ .001) goals reported more belief that people should take care of them-
but not changes in closeness (B ⫽ ⫺0.05, ns). Weekly increases in selves and that their friendship goal made them feel afraid and
compassionate goals predicted increases in closeness (B ⫽ 0.33, confused; they also reported more loneliness and more conflict
p ⬍ .001) and decreases in loneliness (B ⫽ ⫺0.21, p ⬍ .001). The averaged over 10 weeks. These effects were also observed in HLM
Self-Image ⫻ Compassionate interaction did not significantly pre- analyses of the within-person effects of weekly changes in goals
dict changes in closeness (B ⫽ 0.01, ns), or loneliness (B ⫽ ⫺0.04, on weekly changes in feelings, closeness, loneliness, conflict, and
ns). belief in mutual caring and individualistic caring.
Finally, we conducted HLM analyses on the weekly reports of Feelings of closeness and belief in mutual caring mediated the
number of conflicts, mutual caring beliefs, and individualistic effects of compassionate goals on change in perceived available
caring beliefs, entering the goals simultaneously as Level 1 vari- support and trust over the semester. Although the correlational
ables then entering their Level 1 interaction. Increases in compas- nature of the study prevents us from drawing causal inferences, the
sionate goals predicted decreases in conflicts (B ⫽ ⫺0.21, p ⬍
results did not support a meditational model in which goals me-
.001), zero-sum orientation (B ⫽ ⫺0.10, p ⬍ .01), and increases in
diate the effects of closeness and beliefs on change in support and
non–zero-sum orientation (B ⫽ 0.21, p ⬍ .001). Increases in
trust. Thus, the results are consistent with the hypothesis that
self-image goals predicted increases in conflicts (B ⫽ 0.11, p ⬍
compassionate goals cause feelings of closeness and beliefs that
.005), and mutual caring beliefs (B ⫽ 0.08, p ⬍ .04), but did not
people should take care of each other, which in turn predict change
predict change in individualistic caring beliefs (B ⫽ ⫺0.03, ns).
in social support.
The Self-Image ⫻ Compassionate interaction did not significantly
We hypothesized that self-image goals undermine the effect of
predict change in conflict, mutual caring beliefs, or individualistic
compassionate goals on the development of social support and
caring beliefs (Bs ⬍ 0.02, ps ⬎ .70).5
trust. Consistent with this hypothesis, students with high self-
image goals did not significantly increase in trust or support from
Discussion friends and significant others, even if they had compassionate
Study 1 accomplished several aims. First, consistent with the goals; the effects of compassionate goals on change in trust and
view that people who are high in compassionate goals have an perceived support from friends and significant others were signif-
ecosystem perspective regarding relationships, Study 1 showed icant only among students with low self-image goals. Thus, self-
that average compassionate goals are uniquely associated with image goals undermined the otherwise beneficial effects of com-
belief in the interconnectedness of people and all life and with less passionate goals on social support from friends and significant
zero-sum views of success, higher self-compassion and private others, and trust.
self-consciousness, less avoidant attachment style, lower psycho- Self-image goals also attenuated the positive associations
logical entitlement, and higher agreeableness and extraversion. among average compassionate goals, average closeness, and feel-
Consistent with the view that people with self-image goals have an ing clear and connected. Compassionate goals, on the other hand,
ego-system perspective on relationships, self-image goals are attenuated the negative association among self-image goals, aver-
uniquely associated with zero-sum beliefs about success and with age loneliness, feeling afraid and confused, and conflict. These
low self-compassion, public self-consciousness, social anxiety, results provide strong evidence that self-image goals are associated
attachment insecurity, and psychological entitlement. with negative relationship experiences, goal-related affect, and
Second, Study 1 showed that compassionate goals predict in- individualistic caring beliefs and that self-image goals undermine
creases in social support and trust among first semester college
students. Thus, the results are consistent with our hypothesis that 5
When we entered gender as a Level 2 variable, it moderated within-
people who care about the well-being of others create supportive person effects of compassionate goals on conflict (B ⫽ ⫺0.11, p ⬍ .01).
environments for others and for themselves. Befitting students Tests of simple slopes suggest that weekly compassionate goals predict
living away from home for the first time, this effect was greatest lower conflict in women (B ⫽ ⫺0.25, p ⬍ .001) but not in men (B ⫽
for change in perceived available support from friends and small- ⫺0.02, ns). Gender did not moderate within-person effects of compassion-
ate goals on clear and connected affect (B ⫽ 0.01, ns), afraid and confused
est for changes in perceived available support from family. Fur-
affect (B ⫽ ⫺0.02, ns), closeness (B ⫽ 0.05, ns), loneliness (B ⫽ 0.03, ns),
thermore, students with compassionate goals increased in trust
individualistic caring (B ⫽ ⫺0.04, ns), or belief in mutual caring (B ⫽
over the semester; presumably, their efforts to be supportive of 0.07, ns). Gender also did not moderate within-person effects of self-image
others generally met with a positive response, increasing their goals on clear and connected affect (B ⫽ 0.00, ns), afraid and confused
sense that they could safely give to others. The effects of compas- affect (B ⫽ 0.05, ns), closeness (B ⫽ 0.06, ns), loneliness (B ⫽ 0.07, ns),
sionate goals on change in trust and social support were remark- individualistic caring (B ⫽ ⫺0.12, ns), belief in mutual caring (B ⫽ 0.08,
ably robust; when we controlled for eight different sets of indi- ns), or conflict (B ⫽ ⫺0.01, ns).
the beneficial effects of compassionate goals on these outcomes race as White or European American, 4% reported their race as
over a semester. Black or African American, 16% reported their race as Asian or
In Study 1, we could not identify the mechanism by which Asian American, 5% reported their race Latino(a), and 6% selected
compassionate goals increase perceived available support, and other. The racial composition of the sample closely approximated
self-image goals undermine this effect. Although the goals are the racial composition of the incoming freshman class. Of the 62
clearly associated with different relationship experiences, both on roommate pairs who finished the study, 33 (53%) were same-race
average across the semester and as goals change from week to (28 White, 5 Asian), and 29 were mixed-race (10 White and Asian,
week, in Study 1, we could not distinguish between intrapersonal 7 White and Latino(a), 7 White and other, and 5 White and Black).
and interpersonal accounts of these effects. Do students high in Participants ranged in age from 18 years to 22 years (M ⫽ 18.2).
compassionate goals and low in self-image goals simply perceive Modal parental income was greater than $100,000 per year, with 6
increased available support and trust, independent of the reality of participants reporting parental income under $20,000 and 46 par-
their relationships, or do they actually receive more support from ticipants reporting parental income over $100,000.
others? Because we did not measure support transactions in Study
1, we could not evaluate these two processes as accounts for our Procedure
findings.
In groups of 1 to 5, roommate pairs attended a 1.5 hr session to
learn about the study, give their informed consent, complete the
Study 2
pretest survey, and receive instructions for completing the remain-
In Study 2, we addressed this limitation by examining actual ing 22 surveys. All surveys were administered with UM Lessons
social support given and received in a sample of roommates in software. After completing the pretest survey, participants were
their first semester of college. Roommate pairs volunteered to instructed to complete the 21 daily surveys on the Internet. The
participate early in their freshman year. Participants completed daily surveys took about 10 min to complete and roommates were
pretest measures of social support given to and received from their required to complete daily surveys on the same day. Both members
roommate. Then, each roommate completed 21 daily reports of of each dyad received an e-mail reminder with a link to the survey
compassionate and self-image goals for their relationship. Finally, on the days after at least one roommate did not complete a survey.
each roommate completed posttest measures of social support To retain as many participants as possible in the study, we gave
given and received from their roommate. participants up to 28 days to complete the 21 daily surveys. Once
In Study 2 the actor–partner interdependence model (APIM; roommates had completed 21 daily surveys, they completed the
Kashy & Kenny, 2000; Kenny, Kashy, & Cook, 2006) was used to posttest survey, and they were paid for their participation.
examine whether actors’ average goals across the 21 daily reports At pretest and posttest, participants completed measures of
predict changes in actors’ own reports of social support given and social support received from their roommate and given to their
received from pretest to posttest and whether actors’ goals predict roommate as well as questions about demographics (gender, race
changes in their roommates’ (partners’) reports of social support and ethnicity, age, parental income). Additional measures not
received and given. If compassionate and self-image goals operate germane to the goals of the present investigation were also in-
entirely intrapersonally, we would expect actors’ goals to interact cluded.
to predict changes in social support they report receiving from
their partners, but actor’s goals should not predict changes in Measures
partners’ reports of social support received from or given to the
actors. Alternatively, if the effects of compassionate and self- Social support received from and given to the roommate were
image goals operate interpersonally, actors’ goals should interact measured at pretest and posttest with a shortened version of the
to predict changes in partners’ reports of social support received Inventory of Socially Supportive Behaviors (Barrera, Sandler, &
from and given to actors. Ramsay, 1981). Fifteen items assessed the amount of social sup-
port received from the roommate. Participants indicated the fre-
quency with which their roommate provided support during the
Method previous week on a scale from 1 (not at all) to 5 (every day). A
Participants parallel set of 15 items assessed how often participants provided
these same types of support to their roommate during the previous
Participants were 65 first-semester freshmen roommate dyads at week. Both the measure of support received and the measure of
a large Midwestern university who did not know each other prior support given had high internal reliability at both pretest and
to college and who volunteered for a study of goals and roommate posttest (␣s ⫽ .94).
relationships during the fall semester. All dyads were recruited Each of the 21 daily assessments included measures of goals for
within a 1 month time frame; participants had been living together participants’ relationship with their roommate. We modified the
between 4 and 8 weeks when they began participating. Via adver- measure of compassionate and self-image goals used in Study 1 to
tisements in the campus newspaper and flyers, we offered each apply to roommate relationships. All items began with the phrase,
roommate $25 for completing 23 surveys over 4 weeks ($2 for “Today in my relationship with my roommate, I wanted/tried to”
each pretest and posttest and $1 for each daily survey) plus a $25 and were rated on a scale ranging from 1 (not at all) to 5 (always).
bonus for completing all 23 surveys. All 23 surveys were com- Nine items assessed compassionate goals, including “be supportive
pleted by 62 pairs (95%); 46 pairs (71%) were women, and 19 of my roommate,” “be aware of the impact my behavior might
(29%) were men. Sixty-eight percent of participants reported their have on my roommate’s feelings,” “have compassion for my
roommate’s mistakes and weaknesses,” “avoid being selfish or of the total variance was between dyads, and 18% occurred within
self-centered,” “avoid neglecting my relationships with my room- dyads.
mate,” “avoid doing anything that would be harmful to my room-
mate,” “make a positive difference in my roommate’s life,” “avoid Goals and Social Support at Pretest and Posttest and
doing things that aren’t helpful to me or my roommate,” and “be Change From Pretest to Posttest
constructive in my comments to my roommate.” Seven items
reflected self-image goals, including “avoid being blamed or crit- We conducted analyses using APIM, which allows researchers
icized,” “demonstrate my intelligence,” “avoid the possibility of to simultaneously and independently estimate the effect of actors’
being wrong,” “convince my roommate that I am right,” “avoid independent variables on actors’ and partners’ dependent vari-
coming across as unintelligent or incompetent,” “avoid showing ables, with the MIXED command in SPSS. We structured the data
my weaknesses, “ and “get my roommate to do things my way.” with each dyad represented by two lines of data, allowing each
Both scales had high internal consistency each day of the study participant within a dyad to represent both an actor and a partner
(compassionate goals: .91 ⬍ ␣ ⬍ .97, M␣ ⫽ .95; self-image goals: (see Campbell & Kashy, 2002, for sample arrangement of data).
.75 ⬍ ␣ ⬍ .87, M␣ ⫽ .83). We refer to participants as actors and their roommates as partners.
Because participants within dyads were indistinguishable, we
specified a compound, symmetry covariance structure. We com-
Results puted chronic compassionate and self-image goals by averaging
Overview of Analyses across the 21 daily reports.
All outcomes were tested with a two-step hierarchical regres-
Data analyses proceeded in two phases. In Phase 1, we exam- sion. In the first step, we regressed actors’ outcomes on own goals,
ined participants’ chronic goals and interaction as predictors of controlling for partners’ goals. In the second step, we added actors’
participants’ own reported social support received and given at Self-Image ⫻ Compassionate goal product term to the model, and
pretest and posttest and changed from pretest to posttest. We added partners’ Self-Image ⫻ Compassionate goal product term as
simultaneously examined whether participants’ goals and interac- a control, to ensure that effects were due to actors’ goals rather
tion predict their roommate’s reported social support received and than partners’ goals. To test residual change from pretest to post-
given at pretest and posttest and changed from pretest to posttest. test, we entered the pretest outcome variable as a covariate in both
In Phase 2, we examined a path model in which the interaction analyses. In these analyses, an actor effect occurs when actors’
between actors’ compassionate and self-image goals predict goals predict their own outcomes. A partner effect occurs when
change in partners’ support received, which predicts change in participants’ (i.e., actors’) goals predict their roommates’ (i.e.,
partners’ support given, and which in turn predicts change in partners’) outcome. Table 6 shows the unstandardized regression
actors’ support received. coefficients and ts for the associations among actor and partner
Table 4 presents means, standard deviations, and intrapersonal goals and the outcome variables (i.e., support given and received
correlations for all variables. Within participants, support given at pretest, posttest, and change from pretest to posttest).
and received at pretest and posttest correlated moderately to Actor effects. Students with high compassionate goals re-
highly. As in Study 1, participants’ average daily compassionate ported giving and receiving more support at both pretest and
and self-image goals were correlated, although the correlation was posttest. When we tested residual change in support from pretest to
lower in Study 2 (r ⫽ .27) than in Study 1 (r ⫽ .53); consequently, posttest, compassionate goals predicted increased support received
we tested the effects of goals simultaneously in all analyses. Table and given. Self-image goals predicted higher support given but did
5 presents actor–partner (i.e., participant–roommate) correlations not predict received support at pretest. Self-image goals were
for all variables. unrelated to support given and received at posttest and pretest-to-
Both goals varied more between dyads than within dyads. For posttest changes in support given and received.
compassionate goals, 61% of the total variance was between Next, we examined whether participants’ goals interact to pre-
dyads, and 39% occurred within dyads. For self-image goals, 82% dict change in support participants received or gave from pretest to
Table 4
Means, Standard Deviations, and Intrapersonal Correlations for All Variables in Study 2
Variable 1 2 3 4 5 6 M SD
1. Pretest support given — 3.10 0.88

2. Pretest support received .90*** — 3.05 0.90
3. Posttest support given .62*** .60*** — 3.06 0.85
4. Posttest support received .50*** .56*** .86*** — 2.90 0.90
5. Average daily compassionate goals .51*** .43*** .56*** .51*** — 3.34 0.80
6. Average daily self-image goals .30*** .21* .10 .15 .27** — 2.11 0.68
Note. N ⫽ 130 at pretest; N ⫽ 124 at posttest. Chronic compassionate and self-image goals were calculated by averaging across the 21 daily reports.
Support given and received was measured on a scale ranging from 1 (not at all) to 5 (about every day). Chronic self-image and compassionate goals were
measured on a scale ranging from 1 (never) to 5 (always).
*
p ⬍ .05. * p ⬍ .01. * p ⬍ .001.
Table 5
Participant–Roommate Correlations for All Variables in Study 2
Variable 1 2 3 4 5 6
***
1. Pretest support given .59
2. Pretest support received .53*** .48***
3. Posttest support given .48*** .46*** .60***
4. Posttest support received .37*** .34** .42*** .42***
5. Average daily compassionate goals .39*** .39*** .36*** .35*** .38**
6. Average daily self-image goals .16 .13 ⫺.02 ⫺.05 ⫺.01 .17
Note. N ⫽ 130 at pretest; N ⫽ 124 at posttest.

*
p ⬍ .05. * p ⬍ .01. * p ⬍ .001.
posttest. In hierarchical APIM analyses, the Self-Image ⫻ Com- partners give to predict change in support actors receive from
passionate product term did not significantly predict change in pretest to posttest. Figure 4 shows path analyses of the effects of
support received, (B ⫽ ⫺0.05), t(109.15) ⫽ ⫺0.78, R2 ⫽ .00, ns, Compassionate ⫻ Self-Image goals on relationship experiences
or given, (B ⫽ ⫺0.04), t(98.49) ⫽ ⫺0.78, R2 ⫽ .00, ns.6 Thus, the and perceived available social support. We examined the model in
effect of participants’ compassionate goals on the amount of three regression equations, testing each individual path from left to
support they received or gave did not depend on their self-image right, controlling for all variables preceding the path of interest in
goals.7 each equation. As in previous analyses, we arranged the data in a
Partner effects. When participants (actors) had high compas- pairwise structure such that each individual served as both an actor
sionate goals, their roommates (partners) reported receiving and and a partner, used the mixed command in SPSS to account for the
giving more support at pretest and posttest. When we examined nonindependence of roommates, and specified a compound, sym-
residual change from pretest to posttest, actors’ compassionate metry covariance structure because roommates were nondistin-
goals marginally predicted increased support received by partners guishable.
from pretest to posttest, and the goals were unrelated to support Results support our hypothesized model. When participants
given by partners. Actors’ self-image goals were unrelated to (actors) had low self-image goals, their compassionate goals pre-
partner support given and received at pretest and posttest. Exam- dicted greater increases in support received by their roommates
ining residual change from pretest to posttest, actors’ self-image (partners). This increase in partners’ perceived support predicted
goals marginally predicted decreased support received by partners partners giving support to actors, which predicted increased sup-
from pretest to posttest, and the goals were unrelated to change in port received by actors. As shown in Figure 4, the interaction
support given to the partner. significantly predicted change in partner support received from
When we entered the interaction into the regression equation in pretest to posttest ( pr ⫽ ⫺.18, p ⬍ .05). In turn, increases in the
the second block of the hierarchical APIM analyses, it significantly amount of support received by the partner from pretest to posttest
predicted pretest to posttest change in partners’ support received predicted increases in the amount of support given by the partner
(B ⫽ ⫺0.14), t(109.11) ⫽ ⫺2.27, p ⬍ .05, R2 ⫽ .04, and given from pretest to posttest ( pr ⫽ .79, p ⬍ .001). Finally, changes in
(B ⫽ ⫺0.12), t(98.49) ⫽ ⫺2.20, p ⬍ .05, R2 ⫽ .04. Figures 3A support given by the partner predicted increased support received
and 3B depict the predicted values for the interaction between by the actor from pretest to posttest ( pr ⫽ .26, p ⬍ .01).
actors’ goals predicting change in partners’ support received and
given. We computed simple slopes for the effect of partner com- Discussion
passionate goals at high and low levels of self-image goals (1 SD
above and below the mean). For actors low in self-image goals, Consistent with our hypotheses, the results of Study 2 suggest
actor compassionate goals predicted increased partner support that compassionate and self-image goals predict increased social
received (B ⫽ 0.23), t(115) ⫽ 2.76, p ⬍ .01, and partner support support through interpersonal processes, not only intrapersonal
given (B ⫽ 0.18), t(116) ⫽ 2.51, p ⬍ .05. For actors high in processes. Using APIM analyses of roommate pairs, we found that
self-image goals, actor compassionate goals did not significantly participants’ average levels of compassionate and self-image goals
predict change in partner support received (B ⫽ ⫺0.05), t(115) ⫽ over 21 days interacted to predict changes in their roommates’
⫺0.46, ns, or partner support given (B ⫽ ⫺0.06), t(109) ⫽ ⫺0.66, reports of support received from participants and support given to
ns. These results suggest an interpersonal process in which stu- participants. The more participants had compassionate goals, the
dents’ compassionate goals predict increases in the social support more their roommates’ reports of social support received from
their roommates receive and give. participants increased, but only if the participants were low in
6
Path Analyses Following the recommendation of Kenny, Kashy, & Cook (2006), we
report pseudo R2 to demonstrate the variance accounted for by the Self-
Next we tested whether the interaction between actors’ goals Image ⫻ Compassionate product terms in all multilevel models.
predicts change, from pretest to posttest, in support their partners 7
Notably, all actor effects remained unchanged when partner effects
receive, which in turn predicts change in support partners give were removed from analyses. Thus, actor effects were not due to the
from pretest to posttest. Further, we expected change in support number of variables included in the model.
Table 6
Unstandardized Regression Coefficients and ts for the Associations Among Actor and Partner Mean Daily Compassionate and
Compassionate Goals and Support Given and Received at Pretest and Posttest, and Change From Pretest to Posttest in Study 2
Social support received Social support given
Pretest Posttest Change Pretest Posttest Change
Goal B t B t B t B t B t B t
Actors’ compassionate goals

Actors’ support 0.26*** 3.47 0.37*** 4.87 0.26*** 3.64 0.32*** 4.90 0.42*** 6.50 0.29*** 4.45
Partners’ support 0.24** 3.21 0.22** 2.97 0.13† 1.83 0.21** 3.10 0.18** 2.70 0.10 1.56
Actors’ self-image goals
Actors’ support 0.12 1.62 0.05 0.75 0.01 0.12 0.16* 2.50 ⫺0.02 ⫺0.39 ⫺0.09 ⫺1.56
Partners’ support 0.03 0.43 ⫺0.11 ⫺0.53 ⫺0.12† ⫺1.88 0.06 0.85 ⫺0.05 ⫺0.80 ⫺0.07 ⫺1.31
Note. N ⫽ 130 at pretest; N ⫽ 124 at posttest.

†
p ⬍ .07. * p ⬍ .05. **p ⬍ .01. *** p ⬍ .001.
self-image goals. Furthermore, participants’ goals interacted to available social support. In Study 2, goals interacted in a similar way
predict increases in roommates’ reports of social support given to to predict changes over 3 weeks in the amount of support participants’
participants. The more participants had compassionate goals, the roommates reported receiving and giving. Recent evidence that mo-
more roommates’ reports of support given to participants in- tivation and goals can operate outside of conscious awareness raises
creased; this effect was stronger for participants who had low concerns about the validity of explicit self-report measures of goals
self-image goals. Our analyses controlled for roommate’s goals, (Bargh & Williams, 2006; Gillath et al., 2006; Gollwitzer, Bargh,
which also predicted roommate’s support received and given. Elliot, & Dweck, 2005; Wilson & Dunn, 2004; Wirth & Schultheiss,
Results of a path model in which participants’ goals interact to 2006; Wirth, Welsh, & Schultheiss, 2006). The association between
predict increases in roommates’ support received, which in turn one person’s goals and another’s reports of social support received
predict increases in roommates’ support given to participants, and indicate that the findings of Study 1 are not merely due to social
which predict increases in participants’ reports of support received, desirability or other self-report biases; compassionate goals changed
were consistent with the hypothesized interpersonal processes. the experience of another person. Furthermore, the findings suggest
The results of Study 2 increase confidence in the findings of Study that people can accurately report on their goals, at least in this context.
1 by replicating the interactive effects of compassionate and self- The present studies do not rule out the possibility that goals can also
image goals on changes in social support. In Study 1, goals interacted operate implicitly, outside of conscious awareness; in fact, as in other
to predict changes over 10 weeks in participants’ own perceived research on automatic activation of goals, we suspect that compas-
Low Actors’ Self-Image Goals

High Actors’ Self-Image Goals
A B
4 4
Partners’ Residual Posttest Support Received
Partners’ Residual Posttest Support Given
3.5 3.5
3 3
2.5 2.5
Low High Low High
Actors’ Compassionate Goals Actors’ Compassionate Goals
Figure 3. Actor’s compassionate goals predicting scores for partner’s social support (A) received and (B) given
at higher and lower levels of actor’s self-image goals. Means are plotted at 1 SD above and 1 SD below the mean
for self-image goals and compassionate goals.
Figure 4. Path analyses of the effects of Compassionate ⫻ Self-Image goals on relationship experiences and
perceived available social support. All estimates are partial correlations.
sionate and self-image goals can operate both implicitly and explic- want a more supportive relationship, it is probably more helpful to
itly. The present findings however, do increase confidence in our focus on how their own goals can improve their relationships by
measurement of these goals. increasing their roommates’ felt support rather than on how their
In contrast to Study 1, the goals did not interact to predict own supportiveness depends on support they receive.
change in participants’ own reports of social support. As in Study
1, participants’ compassionate goals predicted increases in the General Discussion
social support they received, but unlike Study 1, this effect was not
undermined by participants’ self-image goals. Differences between Many life events disrupt available social support; for the thou-
Studies 1 and 2 may account for this finding. First, the measure of sands of students who move away from home for the first time, the
social support received differed; Study 1 assessed perceived avail- transition to college provides an important example. When such
able support from friends, family, and significant others, whereas disruptions occur, the capacity to rebuild social support may in-
Study 2 assessed social support actually received from roommates. fluence adjustment, with implications for mental and physical
Second, the time between measurements of social support was health. The studies reported here show that people with compas-
10 –12 weeks in Study 1, and 3 weeks in Study 2. sionate goals toward others create a supportive environment, with
This pattern of results is consistent with our hypothesis that upward spirals of support that they give to others and receive from
support recipients try to understand the intentions of support others over time, increasing their interpersonal trust in the process.
providers, specifically, support recipients try to understand These positive effects of compassionate goals are undermined,
whether providers offer support mainly to get something for them- however, among people with self-image goals.
selves or offer support out of genuine concern for the well-being of
the recipient. Further research is needed to determine exactly what Relationship Benefits Associated With Compassionate
cues affect support recipients’ perceptions of support when pro- Goals
viders do and do not have self-image goals. It is possible that
actors with self-image goals give support that is unresponsive to The present studies add to a growing body of research on the
the needs of recipients because they are focused more on what they benefits of giving. Research suggests that people who spend a
want for themselves. Alternatively, it is possible that actors who windfall on others experience an increase in happiness, whereas
are high in self-image goals do not actually give increased support those who spend a windfall on themselves do not (Dunn, Aknin, &
when they are high in compassionate goals, even though they Norton, 2007). Prospective longitudinal studies show that those
report doing so. This possibility could be examined in laboratory who provide care to a spouse live longer and that giving care
studies with objective measures of supportive behaviors. decreases mortality better than receiving care (Brown, Nesse,
In APIM analyses, all participants are treated as both actors and Vinokur, & Smith, 2003). Forgiving partner transgressions (Mc-
partners, reflecting the mutual influence of roommates’ goals on Cullough, Worthington, & Rachal, 1997), responding to destruc-
each other (Kenny et al., 2006). Despite this mutual influence, we tive partner behaviors with constructive behaviors (Rusbult, Ver-
suspect that roommates differ in how they construe the causal ette, Whitney, Slovik, & Lipkus, 1991), and sacrificing immediate
relationship between their own goals and their partners’ support- self-interest to promote the well-being of a partner (Van Lange,
iveness. On one hand, participants could focus on how the support Agnew, Harinck, & Steemers, 1997) can contribute to relationship
they do or do not receive from their roommate causes participants health.
to feel supported and to want to give support to their roommate or Consistent with research demonstrating the benefits of giving,
causes them to feel unsupported and to want to withhold support the present studies showed that students with compassionate goals
from their roommate. In this framing of the relationship, partici- to give support to others experienced increased social support. In
pants simply respond to their roommate’s supportiveness (or lack Study 1, students’ average compassionate goals over 10 weeks
thereof); they are at the mercy of unsupportive roommates. On the predicted increases in perceived available social support from
other hand, participants could focus on their own supportiveness, friends, family, and significant others in the first semester of
which affects their roommates’ feelings of being supported and college. In Study 2, students’ average compassionate goals for
their roommate’s desire to provide support to the participant. their relationship with their roommate predicted increases in social
According to our analyses, both framings represent accurate but support received from roommates over 3 weeks.
incomplete understanding of the reciprocal effects that roommates These studies also clarify how compassionate goals increase
have on each other. Each person’s goal to give support or to social support and trust. In Study 1, we examined intrapersonal
construct desired images affects the experience and intention of the processes and found that when students were high in compassion-
other, with ripple effects for the relationship, potentially creating ate goals they reported believing in mutuality of caregiving, feel-
upward or downward spirals in the relationship. For students who ing clear and connected, feeling close to others and not lonely, and
experiencing less interpersonal conflict. These associations held something in return. Self-image goals undermine the beneficial
both when we examined the effects of goals averaged across 10 effects of compassionate goals. In Study 1, compassionate goals
weeks, and when we examined within-person changes in goals did not predict increased available support from friends or signif-
from week to week. In Study 1, two variables accounted for icant others and did not predict trust unless participants also were
changes in perceived available support and trust: average feelings low in self-image goals. Furthermore, in Study 2, self-image goals
of closeness and beliefs that people should take care of each other. eliminated the positive association between participants’ compas-
In Study 1, we could not determine whether these effects were sionate goals and their roommates’ reports of increased social
entirely intrapersonal, with students’ goals changing their feelings support received and given.
of closeness and beliefs in mutual support, which increase their Previous research suggests that in romantic couples, “invisible”
perceived support and trust, or an interpersonal process of support support that is provided but not perceived by the partner predicts
transaction that affect feelings of closeness and beliefs in mutual partner mood and well-being better than support that is provided
support. and perceived (Bolger, Zuckerman, & Kessler, 2000; Gable, Reis,
In Study 2, we examined interpersonal processes and found that & Downey, 2003). The present results do not directly address this
roommates of students high in compassionate goals and low in issue. However, we propose that the effects of support that pro-
self-image goals reported increases in support received and sup- viders offer and recipients perceive on mood depend on the goals
port given. Path models indicated that participants’ goals inter- of both the provider and the recipient. We predict that recipients’
acted to predict increases in the support their roommates received, well-being and mood are enhanced when they believe providers
which in turn predicted increases in the support their roommates offer support out of concern and caring but may be harmed when
gave, and which in turn predicted increases in the support partic- recipients believe the support is offered out of self-image concerns
ipants received from their roommates. or to obtain something for the self (including staying out of trouble
The results of these studies suggest that people with compas- with the partner). Furthermore, the effects of perceiving that one
sionate goals provide support because they genuinely care about has received support may depend on the goals of the recipient;
the well-being of others. Several individual difference variables recipients who have self-image concerns may react poorly to
assessed in Study 1 support this view; compassionate goals were receiving support because the support suggests that they are needy
associated with spiritual transcendence, specifically, the belief that or inferior (Fisher, Nadler, & Whitcher-Alagana, 1982), whereas
all life is interconnected. People with compassionate goals do not recipients high in compassionate goals but low in self-image goals
view relationships as non–zero sum, with positive outcomes for the may show increases in mood and well-being when they perceive
self being achieved at the expense of others. These people have their partner providing support.
compassion for themselves, as well as others. On weeks that their These results also suggest that the benefits of giving (e.g.,
compassionate goals increased, they reported that their friendship Brown & Brown, 2006; Dunn et al., 2007), sacrifice (Van Lange
goals made them feel clear, loving, and connected to others, that et al., 1997), accommodation (Rusbult et al., 1991), and forgive-
they felt close and less in conflict with others and that they ness (McCullough et al., 1997) for relationship quality may depend
believed that people should take care of each other. These findings on the goals underlying these behaviors. Forgiving because one
suggest that people with compassionate relationship goals do not feels compassion for the mistakes and weaknesses of others may
give support strategically to obtain support for themselves; how- have positive relationship consequences that are undermined when
ever, their belief that people are interconnected and their non–zero- people have self-image goals, such as to appear accepting because
sum view of relationships and caring suggest that they do view one is afraid of losing one’s partner. Similarly, the effects of
giving support as costly to themselves. sacrifice and accommodation on relationship quality may depend
Thus, these studies indicate that compassionate goals do not on the goals underlying this generosity.
reflect altruism, in the sense of costly giving with no benefit to the
self (Batson, 1998). Standard definitions of altruism as costly Implications for the Study of Communal Relationships
giving reflect a zero-sum relationship between self and other;
compassionate goals, on the other hand, reflect a non–zero-sum These results suggest that people can have multiple goals in
perspective. Indeed, students who strive to give support to others communal relationships. As conceptualized by Clark and Mills
reap considerable benefits in social support received, increased (Clark et al., 1986; Mills, Clark, Ford, & Johnson, 2004), commu-
trust, feelings of closeness, low loneliness, and decreased conflict. nal relationships are characterized by rules or norms that people
should respond to the needs of others and expect others to do the
Undermining Effects of Self-Image Goals same for them; in these relationships, people feel obligated or
responsible for meeting the needs of others. Experimental research
People with self-image goals, on the other hand, tend to view on communal relationship orientations has manipulated the desire
relationships as non–zero sum and believe that people should take to have a communal relationship by conveying the impression that
care of themselves, even at the expense of others. They are high in an attractive other was available and anxious to meet new people
public self-consciousness and social anxiety. Study 2 showed that (Clark et al., 1986). As an individual difference, communal rela-
people with self-image goals report giving less support to their tionship orientation taps the general tendency to follow communal
roommate. We hypothesize that when people have self-image rules in relationships (Clark et al., 1987). Thus, operationalizations
goals, they focus more on their own needs and desires than others’ of communal relationships do not clearly distinguish among re-
needs, and they want others to recognize and acknowledge their sponding to the needs of others because people want to demon-
desirable qualities so others will give them what they want. People strate something about the self (e.g., that they are a good friend),
with self-image goals may act supportive mainly, or in part, to get because people want to obtain something (e.g., to have a relation-
ship with an attractive single person), or simply because people home or to disruption of social support resulting from positive life
care about the well-being of others. transitions.
The present studies suggest that people may have either or both
of these goals in communal relationships and that these two types
of goals interact to shape the supportiveness of the relationship
Conclusions
and, perhaps also, relationship quality. Furthermore, they suggest Considerable research has demonstrated the importance of per-
that these goals are associated with distinct emotional experiences. ceived social support for physical and mental health. Because life
Emotion researchers suggest that love evolved to motivate giving events and transitions can disrupt social support, it is important to
to others (Brown & Brown, 2006; Gonzaga, Turner, Keltner, understand how people can foster or impede the social support
Campos, & Altemus, 2006). Study 1 showed that compassionate available to them. The present research suggests that students can
goals are strongly associated with feeling clear, loving, and con- build social support and experience greater closeness and less
nected. Self-image goals, on the other hand, elicit feeling afraid, loneliness in the first semester of college by striving to support
pressured, and confused, as people try to control how they appear others, but only if they are willing to forego the goal of construct-
to others, out of fear that they will not obtain what they want. ing desired self-images. We find these results both important and
As for what people want when they have self-image goals, we promising because they raise the possibility that following disrup-
believe people often adopt self-image goals as a means to gain tions to their social support, people do not need to be at the mercy
acceptance and a sense of belonging or to avoid exclusion. Social of whether social support is perceived to be available; they can
exclusion promotes a desire to connect with others (Maner, DeW- create it by giving it to others.
all, Baumeister, & Schaller, 2007) and, at the same time, it de-
creases prosocial behavior, presumably through a lack of empathy
for others (Twenge, Baumeister, DeWall, Ciarocco, & Bartels, References
2007). Just as social exclusion predicts self-defeating behavior
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
(Baumeister, Twenge, & Nuss, 2002) and impaired self-regulation interpreting interactions. Newbury Park, CA: Sage.
(Baumeister, DeWall, Ciarocco, & Twenge, 2005), self-image Axelrod, R. (1984). The evolution of cooperation. New York: Basic Books.
goals may actually thwart efforts to connect. People with self- Axelrod, R., & Dion, D. (1988, December 9). The further evolution of
image goals may have drawn conclusions from past experience cooperation. Science, 242, 1385–1390.
about how they must appear to others to be accepted or to avoid Bargh, J. A., & Williams, E. L. (2006). The automaticity of social life.
exclusion; they may be unaware of how their own self-image goals Current Directions in Psychological Science, 15, 1– 4.
undermine their efforts to connect. Barrera, M., Sandler, I. N., & Ramsay, T. B. (1981). Preliminary devel-
opment of a scale of social support: Studies on college students. Amer-
ican Journal of Community Psychology, 9, 435– 447.
Limitations Batson, C. D. (1998). Altruism and prosocial behavior. In D. T. Gilbert,
S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology
These studies do not enable us to specify causal relationships (4th ed., Vol. 2, pp. 282–316). New York: McGraw-Hill.
among variables. Although mediation analyses in Study 1 and path Baumeister, R. F. (1982). A self-presentational view of social phenomena.
Psychological Bulletin, 91, 3–26.
analyses in Study 2 were consistent with the view that goals
Baumeister, R. F., DeWall, C. N., Ciarocco, N. J., & Twenge, J. M. (2005).
interact to shape social support, only experimental research can Social exclusion impairs self-regulation. Journal of Personality and
definitely establish the direction of causality. We suspect that the Social Psychology, 88, 589 – 604.
consequences of goals for relationships are reciprocal; compas- Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for
sionate goals may cause increases in closeness, support, and trust, interpersonal attachments as a fundamental human motivation. Psycho-
and increases in closeness, support, and trust may cause compas- logical Bulletin, 117, 497–529.
sionate goals, as long as self-image goals do not intervene to Baumeister, R. F., Twenge, J. M., & Nuss, C. K. (2002). Effects of social
undermine these effects. exclusion on cognitive processes: Anticipated aloneness reduces intelli-
It is unclear whether the results of these studies generalize to gent thought. Journal of Personality and Social Psychology, 83, 817–
827.
other populations. Participants in our studies had recently begun
Bolger, N., Zuckerman, A., & Kessler, R. C. (2000). Invisible support and
college, in most cases moving away from home for the first time. adjustment to stress. Journal of Personality and Social Psychology, 79,
The resulting disruption of social support may be a prerequisite for 953–961.
observing effects of goals; we would not expect compassionate Bowlby, J. (1969). Attachment and loss: Vol. 1. Attachment. New York:
goals to have the same level of beneficial effects in a population Basic Books.
with established and stable sources of social support. Also, com- Bradley, G. W. (1978). Self-serving biases in the attribution process: A
passionate and self-image goals might have different effects on reexamination of the fact or fiction question. Journal of Personality and
other age cohorts, such as older people. Developmentally, the Social Psychology, 36, 56 –71.
issues facing college students, such as establishing independence Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measure-
ment of adult attachment: An integrative overview. In J. A. Simpson &
and identity, differ from the issues facing older people. However,
W. S. Rholes (Eds.), Attachment theory and close relationships. New
research on relationships between people with terminal illness and York: Guilford Press.
their caregivers also indicates that goals to be supportive have Brown, S. L., & Brown, R. M. (2006). Selective investment theory:
beneficial effects for both caregiver and care recipient (Roberts, Recasting the functional significance of close relationships. Psycholog-
Wise, & DuBenske, in press), suggesting that the benefits of ical Inquiry, 17, 1–29.
compassionate goals are not limited to young people first leaving Brown, S. L., Nesse, R. M., Vinokur, A. D., & Smith, D. M. (2003).
Providing social support may be better than receiving it: Results from a Fenigstein, A. (1987). On the nature of public and private self-
prospective study. Psychological Science, 14, 320 –327. consciousness. Journal of Personality, 55, 543–554.
Burton, E., Stice, E., & Seeley, J. R. (2004). A Prospective test of the Fenigstein, A., Scheier, M. F., & Buss, A. H. (1975). Public and private
stress-buffering model of depression in adolescent girls: No support self-consciousness: Assessment and theory. Journal of Consulting and
once again. Journal of Consulting and Clinical Psychology, 72, 689 – Clinical Psychology, 43, 522–527.
697. Fisher, J. D., Nadler, A., & Whitcher-Alagana, S. (1982). Recipient reac-
Buss, D. M., & Kenrick, D. (1997). Evolutionary social psychology. In tions to aid. Psychological Bulletin, 91, 27–54.
D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social Fiske, S. T. (2003). Five core social motives, plus or minus five. In S. J.
psychology (4th ed., Vols. 1 & 2, pp. 982–1026). New York: McGraw- Spencer, S. Fein, M. P. Zannab & J. Olson (Eds.), Motivated social
Hill. perception: The Ontario Symposium (Vol. 9, pp. 233–246). Mahwah,
Campbell, W. K., Bonacci, A. M., Shelton, J., Exline, J. J., & Bushman, NJ: Erlbaum.
B. J. (2004). Psychological entitlement: Interpersonal consequences and Fritz, H. L., & Helgeson, V. S. (1998). Distinctions of unmitigated com-
validation of a self-report measure. Journal of Personality Assessment, munion from communion: Self-neglect and overinvolvement with oth-
83, 29 – 45. ers. Journal of Personality and Social Psychology, 75, 121–140.
Campbell, L., & Kashy, D. A. (2002). Estimating actor, partner, and Gable, S. L., Reis, H. T., & Downey, G. (2003). He said, she said: A
interaction effects for dyadic data using PROC MIXED and HLM: A quasi-signal detection analysis of daily interactions between close rela-
user-friendly guide. Personal Relationships, 9, 327–342. tionship partners. Psychological Science, 14, 100 –105.
Carnelley, K. B., Pietromonaco, P. R., & Jaffe, K. (1996). Attachment, Garner, D. M., Olmstead, M. P., & Polivy, J. (1983). Development and
caregiving, and relationship functioning in couples: Effects of self and validation of a multidimensional eating disorder inventory for anorexia
partner. Personal Relationships, 3, 257–278. nervosa and bulimia. International Journal of Eating Disorders, 2,
Clark, M. S., Mills, J., & Powell, M. C. (1986). Keeping track of needs in 15–34.
communal and exchange relationships. Journal of Personality and So- Gillath, O., Mikulincer, M., Fitzsimons, G. M., Shaver, P. R., Schachner,
cial Psychology, 51, 333–338. D. A., & Bargh, J. A. (2006). Automatic activation of attachment-related
Clark, M. S., Oullette, R., Powell, M. C., & Milberg, S. (1987). Recipient’s goals. Personality and Social Psychology Bulletin, 32, 1375–1388.
mood, relationship type, and helping. Journal of Personality and Social Gollwitzer, P. M., Bargh, J. A., Elliot, A. J., & Dweck, C. S. (2005).
Psychology, 53, 94 –103. Automaticity in goal pursuit. In A. J. Elliot & C. S. Dweck (Eds.),
Cohen, S., & Syme, S. L. (1985). Social support and health. San Diego, Handbook of competence and motivation (pp. 624 – 646). New York:
CA: Academic Press. Guilford Press.
Collins, N. L., & Feeney, B. C. (2000). A safe haven: An attachment theory Gonzaga, G. C., Turner, R. A., Keltner, D., Campos, B., & Altemus, M.
perspective on support seeking and care giving in close relationships. (2006). Romantic love and sexual desire in close relationships. Emotion,
Journal of Personality and Social Psychology, 78, 1053–1073. 6, 163–179.
Collins, N. L., & Feeney, B. C. (2004). Working models of attachment Gosling, S. D., Rentfrow, P. J., & Swann, Jr., W. B. (2003). A very brief
shape perceptions of social support: Evidence from experimental and measure of the Big Five domains. Journal of Research in Personality,
observational studies. Journal of Personality and Social Psychology, 87, 37, 504 –528.
363–383. Greenwald, A. G., & Breckler, S. J. (1985). To whom is the self presented?
Cooley, C. H. (1956). Human nature and the social order. New York: In B. Schlenker (Eds.), The self and social life (pp. 126 –145). New
Schocken. (Original work published 1902) York: McGraw Hill.
Crocker, J. (in press). From egosystem to ecosystem: Implications for Hardin, C. D., & Higgins, E. T. (1996). Shared reality: How social
learning, relationships, and well-being. In H. Wayment & J. Brauer verification makes the subjective objective. In R. Sorrentino & E. T.
(Eds.), Transcending self-interest: Psychological explorations of the Higgins (Eds.), Handbook of motivation and cognition (Vol. 3, pp.
quiet ego. Washington, DC: American Psychological Association. 28 – 84). New York: Guilford Press.
Crocker, J., & Park, L. E. (2004). The costly pursuit of self-esteem. Kashy, D. A., & Kenny, D. A. (2000). The analysis of data from dyads and
Psychological Bulletin, 130, 392– 414. groups. In H. T. Reis & C. M. Judd (Eds.), Handbook of research
Cropley, M., & Steptoe, A. (2005). Social support, life events, and physical methods in social and personality psychology (pp. 451– 477). New York:
symptoms: A prospective study of chronic and recent life stress in men Cambridge University Press.
and women. Psychology, Health, & Medicine, 10, 317–325. Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis.
Crowne, D., & Marlowe, D. (1964). The approval motive: Studies in New York: Guilford Press.
evaluative dependence. New York: Wiley. Kenny, D. A., Korchmaros, J. D., & Bolger, N. (2003). Lower level
Deci, E. L., & Ryan, R. M. (2000). The “what” and “why” of goal pursuits: mediation in multilevel models. Psychological Methods, 8, 115–128.
Human needs and the self-determination of behavior. Psychological Kollock, P. (1993). “An eye for an eye leaves everyone blind”: Coopera-
Inquiry, 11, 227–268. tion and accounting systems. American Sociological Review, 58, 768 –
Dunn, E. W., Aknin, L. B., & Norton, M. I. (2007). From wealth to 786.
well-being: Spending money on others promotes happiness. Unpub- Krull, J. L., & MacKinnon, D. P. (2001). Multilevel modeling of individual
lished manuscript. and group level mediated effects. Multivariate Behavioral Research, 30,
Feeney, B. C., & Collins, N. L. (2001). Predictors of caregiving in adult 41– 62.
intimate relationships: An attachment theoretical perspective. Journal of Lakey, B., Adams, K., Neely, L., Rhodes, G., Lutz, C. J., & Sielky, K.
Personality and Social Psychology, 80, 972–994. (2002). Perceived support and low emotional distress: The role of
Feeney, B. C., & Collins, N. L. (2003). Motivations for caregiving in adult enacted support, dyad similarity, and provider personality. Personality
intimate relationships: Influences on caregiving behavior and relation- and Social Psychology Bulletin, 28, 1546 –1555.
ship functioning. Personality and Social Psychology Bulletin, 29, 950 – Lakey, B., & Cassady, P. B. (1990). Cognitive processes in perceived
968. social support. Journal of Personality and Social Psychology, 59, 337–
Feeney, J. A. (1996). Attachment, caregiving, and marital satisfaction. 343.
Personal Relationships, 3, 401– 416. Lakey, B., Drew, J. B., & Sirl, K. (1999). Clinical depression and percep-
tions of supportive others: A generalizability analysis. Cognitive Ther- Sarason, B. R., Sarason, I. G., & Gurung, R. A. R. (1997). Close personal
apy and Research, 23, 511–533. relationships and health outcomes: A key to the role of social support. In
Lakey, B., McCabe, K. M., Fisicaro, S. A., & Drew, J. B. (1996). Envi- S. Duck (Ed.), Handbook of personal relationships (2nd ed., pp. 547–
ronmental and personal determinants of support perceptions: Three 573). New York: Wiley.
generalizability studies. Journal of Personality and Social Psychology, Sarason, I. G., Pierce, G. R., & Sarason, B. R. (1990). Social support and
70, 1270 –1280. interactional processes: A triadic hypothesis. Journal of Social and
Leary, M. R. (1995). Self-presentation: Impression management and in- Personal Relationships, 7, 495–506.
terpersonal behavior. Madison, WI: Brown & Benchmark. Schlenker, B. R. (1980). Impression management: The self-concept, social
Leary, M. R., & Downs, D. L. (1995). Interpersonal functions of the identity, and interpersonal relations. Monterey, CA: Brooks/Cole.
self-esteem motive: The self-esteem system as sociometer. In M. H. Schlenker, B. R. (2003). Self-presentation. In M. R. Leary & J. P. Tangney
Kernis (Ed.), Efficacy, agency, and self-esteem (pp. 123–144). New (Eds.), Handbook of self and identity (pp. 492–518). New York: Guil-
York: Plenum Press. ford Press.
LeMay, E. P. J., Clark, M. S., & Feeney, B. C. (2007). Projection of Shrauger, J. S., & Schoeneman, T. J. (1979). Symbolic interactionist view
responsiveness to needs and the construction of satisfying communal of self-concept: Through the looking glass darkly. Psychological Bulle-
relationships. Journal of Personality and Social Psychology, 92, 834 – tin, 86, 549 –573.
853. Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the
Lutz, C. J., & Lakey, B. (2001). How people make support judgments: integrity of the self. In L. Berkowitz (Ed.), Advances in experimental
Individual differences in the traits used to infer supportiveness in others. social psychology (Vol. 21, pp. 261–302). New York: Academic Press.
Journal of Personality and Social Psychology, 81, 1070 –1079. Swingle, P. G. (1970). Exploitative behavior in non-zero-sum games.
Maner, J. K., DeWall, C. N., Baumeister, R. F., & Schaller, M. (2007). Journal of Personality and Social Psychology, 16, 121–132.
Does social exclusion motivate interpersonal reconnection? Resolving Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social-
the “porcupine problem.” Journal of Personality and Social Psychology, psychological perspective on mental health. Psychological Bulletin, 103,
92, 42–55. 193–210.
Maslow, A. H. (1968). Motivation and personality. New York: Harper & Tesser, A. (1988). Toward a self-evaluation maintenance model of social
Row. behavior. In L. Berkowitz (Ed.), Advances in experimental social psy-
McCullough, M. E., Worthington, E. L., Jr., & Rachal, K. C. (1997). chology (Vol. 21, pp. 181–227). San Diego, CA: Academic Press.
Interpersonal forgiving in close relationships. Journal of Personality and Tetlock, P. E., & Manstead, A. S. R. (1985). Impression management
Social Psychology, 73, 321–336. versus intrapsychic explanations in social psychology: A useful dichot-
Messick, D. M. (1967). Interdependent decision strategies in zero-sum omy? Psychological Review, 92, 59 –77.
games: A computer-controlled study. Behavioral Science, 12, 33– 48. Thibaut, J. W., & Kelley, H. H. (1959). The social psychology of groups.
Mills, J., Clark, M. S., Erber, R., & Gilmour, R. (1994). Communal and New York: Wiley.
exchange relationships: Controversies and research. Hillsdale, NJ: Erl- Twenge, J. M., Baumeister, R. F., DeWall, C. N., Ciarocco, N. J., &
baum. Bartels, J. M. (2007). Social exclusion decreases prosocial behavior.
Mills, J., Clark, M. S., Ford, T. E., & Johnson, M. (2004). Measurement of Journal of Personality and Social Psychology, 92, 56 – 66.
communal strength. Personal Relationships, 11, 213–230. Uchino, B. N. (2004). Social support and physical health: Understanding
Neff, K. D. (2003). The development and validation of a scale to measure the health consequences of our relationships. New Haven, CT: Yale
self-compassion. Self and Identity, 2, 223–250. University Press.
Piedmont, R. L. (1999). Does spirituality represent the sixth factor of Uchino, B. N., Cacioppo, J. T., & Kiecolt-Glaser, J. K. (1996). The
personality? Spiritual transcendence and the five-factor model. Journal relationship between social support and physiological processes: A
of Personality, 67, 985–1013. review with emphasis on underlying mechanisms and implications for
Pierce, G. R., Sarason, B. R., & Sarason, I. G. (1996). Handbook of social health. Psychological Bulletin, 119, 488 –531.
support and the family. New York: Plenum. Van Lange, P. A. M., Agnew, C. R., Harinck, F., & Steemers, G. E. M.
Raskin, R., & Terry, H. (1988). A principal-components analysis of the (1997). From game theory to real life: How social value orientation
Narcissistic Personality Inventory and further evidence of its construct affects willingness to sacrifice in ongoing close relationships. Journal of
validity. Journal of Personality and Social Psychology, 54, 890 –902. Personality and Social Psychology, 73, 1330 –1344.
Reis, H. T., Clark, M. S., & Holmes, J. G. (2004). Perceived partner Van Lange, P. A. M., Ouwerkerk, J. W., & Tazelaar, M. J. A. (2002). How
responsiveness as an organizing construct in the study of intimacy and to overcome the detrimental effects of noise in social interaction: The
closeness. In D. Mashek & A. P. Aron (Eds.), Handbook of closeness benefits of generosity. Journal of Personality and Social Psychology, 82,
and intimacy (pp. 367–389). Mahwah, NJ: Erlbaum. 768 –780.
Rempel, J. K., Holmes, J. G., & Zanna, M. P. (1985). Trust in close Wilson, T. D., & Dunn, E. W. (2004). Self-knowledge: Its limits, value and
relationships. Journal of Personality and Social Psychology, 49, 95–112. potential for improvement. Annual Review of Psychology, 55, 493–518.
Roberts, L. J., Wise, M., & DuBenske, L. L. (in press). Compassionate Wirth, M. M., & Schultheiss, O. C. (2006). Effects of affiliation arousal
family caregiving in the light and shadow of death. In B. Fehr, S. (hope of closeness) and affiliation stress (fear of rejection) on proges-
Sprecher, & L. Underwood (Eds.), The science of compassionate love: terone and cortisol. Hormones and Behavior, 50, 786 –795.
Research, theory, and applications. New York: Blackwell Publishers. Wirth, M. M., Welsh, K. M., & Schultheiss, O. C. (2006). Salivary cortisol
Rusbult, C. E., & Van Lange, P. A. M. (2003). Interdependence, interac- changes in humans after winning or losing a dominance contest depend
tion, and relationships. Annual Review of Psychology, 54, 351–375. on implicit power motivation. Hormones and Behavior, 49, 346 –352.
Rusbult, C. E., Verette, J., Whitney, G. A., Slovik, L. F., & Lipkus, I. Zimet, G. D., Dahlem, N. W., & Zimet, S. G. (1988). The Multidimen-
(1991). Accommodation processes in close relationships: Theory and sional Scale of Perceived Social Support. Journal of Personality Assess-
preliminary empirical evidence. Journal of Personality and Social Psy- ment, 52, 30 – 41.
chology, 60, 53–78.
Russell, D. W. (2002). In search of underlying dimensions: The use (and Received June 5, 2007
abuse) of factor analysis in Personality and Social Psychology Bulletin. Revision received November 21, 2007
Personality and Social Psychology Bulletin, 28, 1629 –1646. Accepted November 21, 2007 䡲

Creating and Undermining Social Support

Uploaded by

Copyright:

Available Formats

Creating and Undermining Social Support

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Creating and Undermining Social Support

Uploaded by

Copyright:

Available Formats

INTERPERSONAL RELATIONS AND GROUP PROCESSES

Creating and Undermining Social Support in Communal Relationships:

Keywords: social support, goals, self-image, compassion, relationships

Results tions in goals and their interaction predict within-person changes

Compassionate Self-image Compassionate Self-image

Table 1 presents the zero-order correlations between chronic

compassionate goals to predict posttest residual outcomes, we

product term as a second block in a hierarchical regression anal-

posttest, and posttest controlling for pretest.

support to examine change, compassionate goals predicted in-

posttest, but this effect reached statistical significance only for

social support from family.

to predict change in social support from pretest to posttest. In

hierarchical regression analyses, entering the interaction term sig-

nificantly increased the variance explained in posttest social sup-

Mutual caregiving belief

levels of self-image goals (1 SD above and below the mean).

Figure 1 shows the predicted values for compassionate goals

predicting residual posttest social support from friends and from

significant others at higher and lower levels of self-image goals.

For participants low in self-image goals, compassionate goals

predicted increased social support from friends (␤ ⫽ .41, p ⬍ .001)

and significant others (␤ ⫽ .26, p ⬍ .001). For participants high in

Significant other support Family support Friend support Interpersonal trust

Low Self-Image Goals

Residual Posttest Significant Other Support

4.5 nificant, p ⬍ .05.

1. Pretest support given — 3.10 0.88

Note. N ⫽ 130 at pretest; N ⫽ 124 at posttest.

Social support received Social support given

Pretest Posttest Change Pretest Posttest Change

Actors’ compassionate goals

Note. N ⫽ 130 at pretest; N ⫽ 124 at posttest.

Low Actors’ Self-Image Goals

Partners’ Residual Posttest Support Given

You might also like