Murray Once Hurt, Twice Hurtful
-
Upload
ximena-delgado-osorio -
Category
Documents
-
view
231 -
download
0
Transcript of Murray Once Hurt, Twice Hurtful
-
7/27/2019 Murray Once Hurt, Twice Hurtful
1/22
-
7/27/2019 Murray Once Hurt, Twice Hurtful
2/22
Consistent with this idea, interpersonal theorists on the self
argue that self-esteem acts as a sociometer that gauges the risk of
interpersonal rejection (Leary, Haupt, Strausser, & Chokel, 1998;
Leary, Tambor, Terdal, & Downs, 1995). Feelings of self-worth
thus reflect a sense of acceptance, whereas feelings of self-doubt
activate the need for approval and interpersonal connections. Ac-
cordingly, people who feel less valued by others, such as those low
in self-esteem, are likely to be reactive to signs of others approval
because they are motivated to enhance feelings of inclusion
(Nezlek, Kowalski, Leary, Blevins, & Holgate, 1997; Rudich &
Vallacher, 1999). In this way, unfulfilled needs for acceptance
(whether chronic or acute) can sensitize people to social cues,
focusing their mental energies on deciphering the meaning ofevents that might be informative of anothers caring (e.g., Gardner,
Pickett, & Brewer, 2000).
The motivating influence of belongingness needs in shaping
perception and structuring behavior is likely to be particularly
evident in the context of romantic relationships. After all, there
is perhaps no other adult context in which people are so depen-
dent on another person for the satisfaction of their needs
(Braiker & Kelley, 1979; Drigotas & Rusbult, 1992). At rela-
tionship inception, perception of anothers attraction to oneself
is a more potent force in triggering attraction and love than are
considerations of this persons qualities per se (Aron, Dutton,
Aron, & Iverson, 1989; Hazan & Diamond, 2000). As new
relationships develop, moreover, intimates actively monitor
their partners behaviors for signs of selflessness and, thus,reason to trust in their partners willingness to respond to their
needs (Fraley & Shaver, 2000; Holmes & Rempel, 1989;
Wieselquist, Rusbult, Foster, & Agnew, 1999). Even in well-
established marriages, intimates still seek their partners admi-
ration and approval, reporting that they want their partner to see
them much more positively than they see themselves (Murray,
Holmes, & Griffin, 2000).
To find this sense of acceptance, low and high self-esteem
people alike need to believe that others see qualities in them worth
valuing (Leary & Baumeister, 2000). For instance, dating and
married intimates report feeling more accepted and loved by their
partner when they believe their partner sees them more positively
on specific interpersonal qualities (Murray, Holmes, Griffin, Bel-
lavia, & Rose, 2001). In other words, Sally is not likely to believe
that Harry will always love her if she has trouble pinpointing
specific qualities in her that he values, especially ones he could not
easily find in another (Tooby & Cosmides, 1996). Accordingly, in
the present study, we equate the activation of belongingness needs
with the perception that the partner sees relatively few positive
qualities in the self. Conversely, we equate the relative satiation of
belongingness needs with the perception that the partner sees many
positive qualities in the self.1
Unfortunately, however, attentiveness to a partners behaviors
need not always generate accurate insights into a partners true
sentiments toward oneself. For instance, people with low self-
esteem have difficulty finding evidence of a partners acceptance
in even the most accommodating behavioral realities (despite the
fact that they seem strongly motivated to find signs of acceptance
and caring). In dating and marital relationships, people with low
self-esteem underestimate how positively their partner sees them
on specific traits (Murray, Holmes, & Griffin, 2000) and even
underestimate how much their partner loves them (Murray et al.,
2001). People who are chronically sensitive to rejection (and thus
lower in self-esteem) also underestimate their partners satisfaction
and commitment (Downey & Feldman, 1996).
1 In drawing these parallels, we are not trying to argue that feeling
understood by a partner is unimportant to felt acceptance. In fact, Reis
and Shavers (1988) process model of intimacy is based on the assumption
that perceptions of a partners caring depend on the perception of being
understood. Similarly, Swann, Hixon, and De La Ronde (1992) reported
that low self-esteem people report greater intimacy in their marriage when
their partner sees them as negatively as they see themselves. From our
perspective, however, people need to be able to pinpoint particular reasons
why their partner is likely to love and value them before they are likely
to seek a partners understanding or validation of their more negative
qualities.
Figure 1. A situational analysis of felt security regulation.
127PERCEIVED REGARD AND MARITAL INTERACTIONS
-
7/27/2019 Murray Once Hurt, Twice Hurtful
3/22
Seeking Acceptance With a Chronically
Activated Sociometer
What are the likely consequences of more or less positive,
chronic perceptions of a partners regard for the meaning that
might be gleaned from daily relationship events (Path A in Figure
1)? People who feel less positively regarded by their partner mayapproach their relationship in a hypothesis-testing fashion, scruti-
nizing their partners behavior for information about caring. In
scanning the available behavioral data, they may hope to soothe
their own insecurities by finding conclusive evidence that their
partner really does value them. However, they are not likely to be
all that evenhanded in this search. Instead, they may process
information in risk-averse ways, too ready to perceive and gener-
alize from signs of rejection and too hesitant to trust signs of
acceptance. Why might such a seemingly counterproductive asym-
metry emerge?
Feelings of uncertainty and consequent concerns about making
incorrect inferences are thought to result in people processing
information in a cautious or risk-averse fashion (Holmes & Rem-
pel, 1989; Taylor, 1991). That is, people may become more hes-itant to believe what they hope to be true as the costs of being
wrong in this inference increase. Even though detecting signs of
rejection might hurt in the short term, such sensitivity may better
protect intimates who feel less valued by their partner against the
greater hurt of inferring acceptance, risking attachment, and then
later reaching the all the more hurtful conclusion that they were
never really accepted after all. In this way, the combination of
relatively negative chronic expectations and risk aversion may
prime intimates who are looking for signs of acceptance to over-
interpret and internalize signs of rejection as they monitor their
partners day to day behaviors.
On the other side of the coin, confident expectations that a
partner sees positive qualities in the self may inoculate intimatesagainst all but the most obvious signs of a partners rejection in
day to day social interactions. For people who feel more positively
regarded, needs for belongingness are likely to be largely satiated
(and thus quiescent). Rather than looking for signs of a partners
approval, such intimates may instead approach their relationship
with an eye toward confirming and maintaining benevolent expec-
tations about a partners caring. When intimates feel valued, then,
a partners positive behaviors may be easily trusted and assimi-
lated to their expectations, and negative behaviors may be more
easily discounted and consequently may hurt less. In fact, inti-
mates who chronically feel more valued might actively compen-
sate for seemingly threatening events by exaggerating or embel-
lishing just how much their partner values and loves them (see
Murray, 1999, for a review of such motivated cognitive processes).In summary, chronic perceptions of a partners regard for the
self may set the activation threshold or set point of a relationship-
specific sociometer at high or low levels. Differing sensitivities to
rejection may then result in intimates processing day to day events
and partner behaviors in ways that are likely only to reinforce
chronic feelings of being more or less valued. Consistent with this
hypothesis, people with lower self-esteem (and thus less confi-
dence in their partners regard) overinterpret their dating partners
(hypothetical) negative moods, seeing them as symptomatic of a
partners ill feelings toward the self (Bellavia & Murray, in press).
They also react to experimentally induced signs of a partners
irritation by anticipating rejection (Murray, Rose, Bellavia,
Holmes, & Kusche, 2002). Similarly, dating intimates who are
high on attachment-related anxiety interpret a partners hypothet-
ical (Collins, 1996) and actual misdeeds in suspicious ways that
are likely to exacerbate hurt feelings (Simpson, Rholes, & Phillips,
1996).
The Situational Regulation of Dependency
Thou has not half the power to do me harm as I have to be hurt.
William Shakespeare, Othello
How might chronic feelings of being more or less positively
regarded affect peoples capacity to maintain feelings of connec-
tion to the partner in situations in which they are feeling acutely
hurt or vulnerable (Path B in Figure 1)? Sustaining a sense of felt
security in a partners continued availability and responsiveness
seems to require an inferential leap of faith, putting the best
possible spin on the available evidence (particularly when that
evidence is negative).
For instance, people in committed dating relationships benevo-
lently misconstrue their partners attraction to others (Simpson,Ickes, & Blackstone, 1995). People in satisfying marriages also
generously attribute their partners transgressions to features of the
situation (Bradbury & Fincham, 1990), and they inhibit inclina-
tions to respond in kind to a partners misdeeds and, instead,
respond constructively (Rusbult et al., 1991). They also see virtues
in their partner that are not apparent to their friends (Murray,
Holmes, Dolderman, & Griffin, 2000) or their partner himself or
herself (Murray, Holmes, & Griffin, 1996). And people in more
stable, satisfying marriages see strengths in their partner that they
do not see in the partners of others (Rusbult, Van Lange, Wild-
schut, Yovetich, & Verette, 2000; Van Lange & Rusbult, 1995).
However, the capacity to defend the relationship in the face of
threat appears to be constrained by the perceived risks of suchgenerosity (Murray, Holmes, MacDonald, & Ellsworth, 1998;
Murray, Holmes, & Griffin, 2000; Murray et al., 2001). In forgiv-
ing Sallys transgression, for instance, Harry protects a sense of
felt security but leaves himself especially vulnerable to future hurt
if his attributional charity proves to be unfounded. It would not be
surprising, then, if Harry only forgave Sally when he felt confident
of her reciprocated tolerance. Supporting this dependency regula-
tion hypothesis, dating and married intimates are more likely to see
the best in their partners traits when they feel confident that their
partner also sees the best in them (Murray, Holmes, & Griffin,
2000) and when they feel more loved by their partner (Murray et
al., 2001).
In specific situations, then, the tendency to read rejection and
hurt into daily events may make it difficult for intimates whochronically feel less valued by their partner to respond construc-
tively to difficulties. Instead, they might react to the acute pain of
rejection with anger and by taking the defensive step of actively
distancing from the source of the hurtthe partner or relationship.
After all, devaluing the partner, lashing out behaviorally, or re-
ducing feelings of closeness likely all function to lessen the acute
threat to the self posed by feeling hurt or rejected (Murray et al.,
1998). In a sense, then, intimates who chronically feel less valued
may settle for a sense of safety or felt security that comes from
avoiding situations that put the self at risk for further harm a kind
of prevention motivation (see Higgins, 1996).
128 MURRAY, BELLAVIA, ROSE, AND GRIFFIN
-
7/27/2019 Murray Once Hurt, Twice Hurtful
4/22
Supporting this hypothesis, more anxiously attached women
display greater anger toward their partner in situations in which
their partner may not have been as responsive as they hoped
(Rholes, Simpson, & Orina, 1999). Women who chronically an-
ticipate rejection in relationships also behave more negatively
toward their partner during conflicts (Downey, Freitas, Michaelis,
& Khouri, 1998). Low self-esteem people also respond to inducedanxieties about rejection by self-protectively derogating their part-
ner (Murray et al., 1998; Murray et al., 2002). To return to the
sentiment from Othello, intimates who chronically feel less valued
by their partner may react to acute feelings of vulnerability and
hurt by distancing and thus giving their partner less power to hurt
them.
The idea that intimates who chronically feel less valued react to
acute feelings of rejection by finding fault in their partner and
engaging in overtly rejecting behaviors may seem inconsistent
with a basic tenet of the sociometer model. Namely, feelings of
rejection are thought to function as an alarm system that activates
compensatory processes and approach behaviors aimed at securing
acceptance (Leary et al., 1995; Williams, Cheung, & Choi, 2001;
Williams & Sommer, 1997). Following threatening feedback abouttheir intellectual abilities, for instance, people with low self-esteem
are liked more by strangers in a novel interaction than are people
with high self-esteem, suggesting that lows were more worried
about rejection and took steps to avoid this outcome (Heatherton &
Vohs, 2000; Vohs & Heatherton, 2001).
Given this evidence that rejection concerns activate approach
motivations, why would we expect those intimates who feel less
positively regarded and thus are in most need of acceptance to
think and behave in ways that are likely to lead to a partner s
irritation and disinterest? Interactions with an intimate partner are
unique in that the partner is both the source of rejection and the
desired target of approach behaviors. In routine social interactions,
however, people need not seek acceptance from someone whoappears to be rejecting. Instead, they may readily alleviate feelings
of rejection by approaching novel others (as they did in the
research described above). In romantic relationships, people may
not have this luxury, as they are often caught in the position of
feeling hurt by the very person whose acceptance they most need.
It is for this reason that we believe that a sense of acceptance needs
to be secured in romantic relationships before intimates are likely
to be willing to respond to acute feelings of hurt and vulnerability
by behaving constructively, drawing closer, and thus approaching
the source of that hurt. When people question their partners
acceptance, then, even their best attempts to elicit reassurance or
comfort might be cloaked in blame and criticism of the partner.
For people who feel more valued in their partner s eyes, resilient
expectations of acceptance may lessen the momentary sting ofrejection in ways that allow them to put affirming the relationship
ahead of defending against a perceived slight or hurt to the self. In
response to feeling hurt by a badly behaving partner, for instance,
such people may more readily counteract such specific rejection
experiences by reminding themselves of their partners past good
will and caring or by drawing on the protective resource of feeling
generally accepted. As a result, people who feel more positively
regarded may more readily compensate for signs of difficulty by
drawing closer, emphasizing the strengths of their partner and
relationship. In a sense, intimates who feel more valued may find
a sense of felt security in their relationship through approach
behaviors aimed at enhancing the value of the relationshipa kind
of promotion motivation (see Higgins, 1996).
Consistent with the idea that felt acceptance fosters generosity,
unconsciously primed thoughts of security and acceptance dimin-
ish peoples normal tendencies to derogate out-group members
(Mikulincer & Shaver, 2001) and increase peoples desire to seek
support from others in dealing with a personal crisis (Pierce &Lydon, 1998). Similarly, consciously activated feelings of being
unconditionally accepted reduce peoples tendency to disparage
others in response to a threat to the self (Schimel, Arndt,
Pyszczynski & Greenberg, 2001).
Summary of the Hypotheses
The current daily diary study presents the first empirical test of
how feeling more or less positively regarded in a partners eyes
shapes the day to day regulation of felt security and thus structures
the momentary feelings, thoughts, and behavioral interactions of
married couples.
We hypothesized that the chronic activation of belongingness
needs would predispose intimates to actively but cautiously hy-pothesis test and, consequently, read too much into negative but
not positive events in their relationship (the moderation of Path A).
Accordingly, we expected people who felt less positively regarded
in their partners eyes to read more into potentially threatening
events, such as conflicts or a partners bad moods or transgres-
sions, than people who felt more valued. Specifically, we expected
intimates who felt less valued to feel less accepted, more hurt and
rejected, and more anxious about their partners acceptance on
days after they detected reasons to doubt their partner. However,
we expected intimates who felt more positively regarded to more
readily compensate for such perceived slights and difficulties on
one day by reaffirming their partners love and acceptance on
subsequent days.
As intimates who are operating at a chronic inclusion deficithave greater needs for a partners acceptance, acute feelings of hurt
and vulnerability should be particularly painful and, thus, activate
the need to protect the self against further hurt. Accordingly, we
expected people who felt less positively regarded to respond to
feelings of vulnerability by rejecting their partner in advance and,
thus, devaluing what they fear they might lose (the moderation of
Path B). Specifically, we expected people who felt less valued in
their partners eyes to respond to feeling hurt, rejected, or anxious
about a partners acceptance one day by engaging in criticizing,
cold, demanding, and otherwise negative behaviors on subsequent
days. Ironically, then, people who feel less valued may actually
ensure that their partner does evaluate them more negatively, as we
expected the partners of such intimates to report greater feelings ofannoyance and frustration with the partner on days after such
events. However, we expected people who felt more positively re-
garded to more readily resist the temptation to lash out in response to
vulnerability. Instead, we expected these intimates to actively draw
closer to their partner on days after they felt more hurt or rejected,
thereby reaffirming feelings of acceptance and belongingness.
Method
Overview
This research was designed to examine how chronic perceptions of a
partners regard affect the way intimates interpret and then respond to daily
129PERCEIVED REGARD AND MARITAL INTERACTIONS
-
7/27/2019 Murray Once Hurt, Twice Hurtful
5/22
events in marriage. At an initial laboratory session, both members of each
couple completed a background questionnaire tapping perceptions of their
partners regard for them and perceptions of themselves, their partner, and
their relationship. Both members of the couple then completed a 15-min
daily experiences record (DER) each day for 21 days. Each DER asked
participants first to indicate which positive and negative self- and
relationship-relevant events had occurred that day and then to describe
their feelings and thoughts about themselves, their partner, and theirrelationship that day.
Participants
One hundred seventy-three couples involved in marital or cohabiting
relationships participated in a study on Daily Experiences in Marriage
held at the University at Buffalo, State University of New York. Nineteen
couples were not included in data analysis because they completed fewer
than 14 diaries (12 couples), because they had language difficulties (5
couples), or because they provided suspicious records (2 couples). Of the
remaining 154 couples, 152 couples were married, and 2 couples were
cohabiting. Their mean age was 34.4 years. The average duration of their
marriage was 7.6 years. The majority of the couples (89) had at least one
child from the current relationship. The sample was fairly well educated(44% had a high school diploma, 42% had a college/university degree, and
14% had a graduate/professional degree) and in the middle to upper class
(33% had a combined income less than $40,000, 62% had a combined
income between $40,000 and $100,000, and 5% had a combined income
greater than $100,000). Each couple received $90 payment and a chance to
win $100 in a lottery for participation.
Procedure
In recruiting our sample, we placed advertisements promoting the study
in the classified sections of local newspapers, in public service announce-
ments on local radio stations, and on community bulletin boards in grocery
stores, churches, bars, and restaurants. Interested couples contacted our
laboratory by telephone and were screened for their eligibility (i.e., at
least 18 years of age, married 15 years or less or cohabiting at least 2 but
no more than 15 years, and living in the same residence, i.e., not commut-
ing). Couples who met these criteria were then scheduled for a 1.5- to 2-hr
orientation session at our laboratory.
At this orientation session, the research assistant first seated the mem-
bers of the couple at separate desks and then asked each participant to
complete the background questionnaire package. The research assistant
then introduced the daily diary form and described the procedures for
completing it. Specifically, the research assistant emphasized that partici-
pants should begin completing their diaries the following day, that they
should complete the diaries before going to bed, that they should return
their diaries biweekly by mail on specified days, that their responses were
anonymous and confidential, that they should not discuss their diaries with
their partner, and that if they happened to miss a day, they should leave that
record blank.
Both members of the couple then completed a practice diary describing
the events and feelings they had experienced in their relationship that day
(prior to arriving at the lab). The research assistant then answered any
questions that participants had about the format of the diary or about the
wording or meaning of any individual item. The couple then set up a
follow-up appointment for at least 3 weeks after the orientation session.
Both members of the couple left the laboratory with a personalized daily
diary package that contained an instruction sheet, a calendar specifying the
biweekly dates for submitting the diaries, 22 numerically coded diary
records (1 record was provided as a spare), and addressed, stamped, and
dated return envelopes.2
At the 45- to 60-min follow-up session, both members of the couple
completed measures tapping their self-, partner, and relationship evalua-
tions. They then completed a further questionnaire assessing their experi-
ences in completing the DERs. The research assistant then described the
broad purposes of the study to participants, paid them, and thanked both
members of the couple for their participation.
Background Measures
The first page of each background questionnaire asked participants forbasic demographic information (i.e., gender, age, ethnicity, highest level of
education attained, annual salary). The remainder of the questionnaire
contained measures of self-esteem, perceptions of the partners regard,
perceptions of the partner, and overall evaluations of the relationship
(among other measures). We describe below only those measures relevant
to the aims of the current article.
Global self-esteem. Rosenbergs (1965) 10-item measure ( .88)
assessed global self-evaluations (e.g., I feel that I am a person of worth,
at least on an equal basis with others). Participants responded on 7-point
scales (1 strongly disagree, 7 strongly agree).
Perceptions of the self. This 20-item interpersonal qualities scale
(Murray, Holmes, & Griffin, 1996; .80) contained positive and
negative attributes from the interpersonal circle, a model based on the
primary dimensions of warmth hostility and dominancesubmissiveness
(see Murray et al., 1996, for further details). Example attributes includekind and affectionate, open and disclosing, responsive to my needs,
tolerant and accepting, understanding, patient, warm, critical and
judgmental, controlling and dominant, thoughtless, distant, com-
plaining, moody, and irrational. Participants rated how well each of
the traits described themselves on a 9-point scale (1 not at all charac-
teristic, 9 completely characteristic). In computing an overall score, we
reverse scored negative traits, such that higher scores represented more
favorable perceptions.
Perceptions of the partners regard for the self. Participants also
described how they thought their partner saw them on the identical 20
interpersonal qualities to index the extent to which they felt valued when
they imagined themselves through their partners eyes ( .85).
Perceptions of the partner. Participants also described their partner on
the same 20 attributes to index their perceptions of their partner s basic
goodness and value ( .88).
Satisfaction. This four-item scale (Murray, Holmes, & Griffin, 1996;
.92) assessed global evaluations of the relationship (e.g., I am
extremely happy with my current romantic relationship, I have a very
strong relationship with my partner, I do not feel that my current
relationship is successful, My relationship with my partner is very
rewarding, i.e., gratifying, fulfilling). Participants responded to these
items on 9-point scales (1 not at all true, 9 completely true).
The DERs
Each DER contained two sections. The first section asked participants to
complete a 103-item event inventory, checking off those positive and
negative events that had occurred that day. General categories of events
included successes or failures at work, interactions with the spouse, man-
aging household and family responsibilities, interactions with friends and
extended families, and interactions with children. The second section asked
participants to complete a 54-item feelings inventory, rating how much of
each emotion they experienced that day. Emotional reactions centered
around self-evaluations, perceptions of the partners regard for the self,
perceptions of the partner, and overall evaluations of the relationship. We
describe below the categories of events and feelings relevant to the current
article.
2 Participants also completed separate open-ended weekly diaries in
which they wrote about the best and worst events that happened in their
relationship that week.
130 MURRAY, BELLAVIA, ROSE, AND GRIFFIN
-
7/27/2019 Murray Once Hurt, Twice Hurtful
6/22
Own negative behaviors. This five-item measure indexed how often
participants described themselves as engaging in negative behaviors to-
ward their partner that day (i.e., criticized or complained about my
partner, insulted my partner, ignored my partner, was inconsiderate
or selfish, snapped or yelled at my partner). For this (and all following)
behavior composites, we assigned a 1 to checked events and a 0 to
nonchecked events and then summed these scores to provide a continuous
index of the behavior in question on each day.Partners (perceived) negative behaviors. This seven-item measure
indexed how often participants perceived their partner as engaging in
negative behaviors toward them on each day (i.e., partner was irritated or
angry with me, partner criticized or complained about me, partner
insulted me, partner ignored me, partner was inconsiderate or selfish,
partner snapped or yelled at me, partner embarrassed or made fun of me
in public).
Own positive behaviors. This six-item measure tapped how often par-
ticipants described themselves as engaging in positive behaviors toward
their partner each day (i.e., made time to be with my partner, did
something thoughtful for my partner, shared private thoughts with my
partner, changed my behavior for partner, told my partner I loved
him/her, forgave my partner).
Partners (perceived) positive behaviors. This 10-item measure tapped
how many positive behaviors participants perceived in their partner each
day (e.g., partner told me he/she loves me, partner praised me, partner
was physically affectionate, partner did something thoughtful for me,
partner listened to me, partner made time to be with me).
Conflict. This two-item measure assessed the occurrence of overt
conflict between the spouses each day (e.g., we had a minor disagree-
ment, we had a serious argument).
Mood. This six-item measure ( .82) tapped daily mood and self-
evaluation (i.e., unsure of myself, anxious, sad/depressed, unlik-
able, lonely, I wanted to be left alone). Participants rated their
experience of these (and all following) feelings on 7-point scales (1 not
at all, 7 especially). Responses to each item were averaged (for this and
all following daily feelings measures) to create overall scores for each day.
Felt rejection/hurt. This six-item measure ( .91) tapped how
rejected or hurt by their partner participants felt each day (i.e., rejected orhurt by partner, partner doesnt understand me, partner wasnt there for
me, partner was angry with me, partner was irritated with me,
partner doesnt really care what I think).
Felt acceptance. This seven-item measure ( .94) tapped how
accepted, validated, or affirmed participants felt by their partner each day
(e.g., partner loves me, partner accepts me as I am, partner sees the
best in me, partner overlooks my faults, comforted or reassured by
partner, partner is proud of me).
Anxiety about acceptance. This five-item measure ( .86) tapped
feelings of concern about the continued availability of the partners accep-
tance and love each day (i.e., worried about disappointing partner,
unsure whether partner is happy in our relationship, partner is pulling
away from me, partner is bored with me, I care more about this
relationship than my partner).
Anger toward partner. This two-item scale ( .88) tapped feelings
of anger or irritation toward the partner each day (i.e., irritated or annoyed
with my partner, angry with my partner).
Perceptions of partners traits. This four-item scale ( .84) tapped
more global negative evaluations of the partner, focusing on the extent to
which participants thought their partner was being overly demanding,
clingy, or selfish each day (i.e., my partner is selfish, my partner is too
dependent on me, my partner is taking me for granted, my partner nags
me too much).
Closeness. This four-item scale ( .87) tapped how close, con-
nected, and happy intimates felt in their relationship each day. Two of these
items asked participants to rate how in love with their partner they were
and how happy they were in their relationship using the 7-point scale
described above. The remaining two items asked participants to describe
their overall evaluations of their relationships that day (1 terrible, 7
terrific) and to describe how close or connected they felt to their partner on
seven progressively overlapping circles (Aron et al., 1989).
In combination, the above daily indices allow us to explore the day to
day causal links depicted in Figure 1. Using the indices labeled own
negative behavior and felt rejection, for instance, we can link Sallys
self-professed rejecting behavior to Harrys subsequent feelings of hurt orvulnerability (Path A), and we can link Sallys own feelings of hurt to her
subsequent negative behaviors (Path B). Using the indices labeled own
positive behavior and felt acceptance, we can link Sallys self-described
positive behaviors to Harrys subsequent feelings of acceptance (Path A),
and we can link Sallys own feelings of acute acceptance to her subsequent,
self-described positive behavior toward her partner (Path B). Using the
indices labeled partners perceived negative behaviors and anxiety over
acceptance, we can link Sallys perceptions of Harrys negative behaviors
to her subsequent feelings of vulnerability (Path A), and we can link Sallys
acute feelings of anxiety to Harry subsequently describing her as engaging
in negative behaviors (Path B).
Results
We first examine whether perceptions of a partners regard forthe self govern how hurt or rejected intimates feel in response to
negative events in their relationship (the moderation of Path A).
We then examine whether perceptions of a partners regard also
control how intimates respond to feeling hurt and rejected by their
partner (the moderation of Path B). Such hypotheses postulate
cross-level interactions: Within-person effects, such as the link
between the perception of conflict on one day and feelings of
rejection on the next, are moderated by a between-persons effect,
feeling valued by ones spouse.
Testing cross-level hypotheses requires multilevel data-analytic
procedures (Kenny, Kashy, & Bolger, 1998; Nezlek, 2001). To
simplify things for a moment, imagine that our data structure was
reduced and we just had mens responses to the DERs each dayfor 21 days. This data structure would contain two levels of
analysis (day and person), and we would specify equations repre-
senting the relations among variables at each level of analysis.
Level 1 equations specify the links between variables at a daily
level and yield slope and intercept coefficients to index these
relations (e.g., Do men feel more hurt on days after high conflict
days?; a Level 1 slope coefficient). Level 2 equations specify the
relation between these within-person coefficients and between-
persons variables (e.g., Is the tendency to feel more rejected on
days after high conflict days weaker for men who feel more
positively regarded?i.e., are the Level 1 slopes smaller for men
who believe they are more positively regarded?). Because we
have daily diary reports on both members of each couple, we have
a three-level data structure: Individuals (classified by gender) arenested within couple, and day is nested within individuals. This
allows us to test whether the observed relations at the daily level
and the cross-level relations vary across gender.
Using the multivariate feature of the multilevel modeling pro-
gram MlwiN (Goldstein et al., 1998), we modeled our data as a
three-level nested structure with within-person across-day effects
making up the lowest level, between-persons effects making up the
second level, and a variable representing gender within couple
making up the highest level. This approach is essentially identical
to the data structure used in the classic studies by Barnett and
colleagues (e.g., Barnett, Marshall, Raudenbush, & Brennan, 1993;
131PERCEIVED REGARD AND MARITAL INTERACTIONS
-
7/27/2019 Murray Once Hurt, Twice Hurtful
7/22
Barnett, Raudenbush, Brennan, Pleck, & Marshall, 1995; Rauden-
bush, Brennan, & Barnett, 1995), except that it exchanges the use
of multiple dummy variables to indicate gender for the use of a
multivariate command that controls how effects are estimated
simultaneously for men or women within a couple.3 This approach
simultaneously estimates two regression equations, one for
women, and one for men, controlling for the interdependencebetween measures taken on two individuals within a dyad. The
multivariate approach instantiated within MlwiN also allows
straightforward tests of the equality of coefficients across men and
women (i.e., tests of gender differences).
The first stage of analysis required assessing whether within-
person effects of interest (i.e., Level 1 slopes and intercepts) varied
significantly across people. For example, significant variation in
the slopes between conflict yesterday and felt rejection today
would indicate that such slopes (computed for each person) differ
significantly between persons. Such a finding then leads to a
search for variables at the level of the person that might moderate
the Level 1 effects. Significant variation in the within-person
coefficients thus justifies further examination of whether between-
persons effects, such as feeling valued by a partner, moderate thesize of the within-person effects.4 A prototype set of equations for
testing this condition is below:
YWt B0W B1WYt1 B2WCt1 v0W uW (1)
YMt B0M B1MYt1 B2MCt1 v0M uM. (2)
Equation 1 represents the effects for women (subscript W);
Equation 2 represents the effects for men (subscript M). In each
case, we predicted the value of the dependent variable (Y) on a
given day for a given couple from an average level term (B0, the
intercept that varies across individuals and is thus a random
coefficient); the value of Yon the previous day (the coefficient B1
represents a fixed effect, essentially the average within-personslope across all individuals); the value of the independent variable,
such as conflict, on the previous day (B2
, also a fixed effect in this
preliminary model); an error term (v0
) that reflects how much each
persons average deviates from the overall average; and an error
term (u) that reflects each persons daily deviation from his or her
own mean on Y. The within-person variables representing the
lagged stability effect (Yt1) and the independent variable (Ct1)
are centered around each persons mean for that variable. Signif-
icant coefficients for such variables are thus interpreted in terms of
the effects of a person being high or low on a given day relative to
his or her own mean for that variable across days.5
When these equations were initially fit as fixed effects, a devi-
ance statistic was computed that represented the badness of fit of
the model to the data. In a second step, the model was fit with theindependent variable (e.g., conflict) set as a random coefficient;
that is, it was allowed to vary across individuals. We then com-
pared the deviance of the fixed effects model with that of the
random effects model. The deviance should decrease significantly
from the fixed to random effects model if the slopes have signif-
icant random variance across people (a seven degrees of freedom
test).6
If the deviance tests reveal a significant fit improvement with
the random slopes, we then add moderating variables to the equa-
tions above to try to capture the between-persons variation in the
Level 1 slopes or intercepts. A prototype set of resulting equations
is below and includes one random intercept (B0
) and one random
slope (B2):
YWt B0W B1WYt1 B2WCt1 B3WZ
B4WZC v0W v2W uW (3)
YMt B0M B1MYt1 B2MCt1 B3MZ
B4MZC v0M v2M uM. (4)
In this next stage of analysis, we predicted the value of the
dependent variable (Y) on a given day from the randomly varying
intercept term (B0
), the fixed effect of Y on the previous day (B1
),
the random effect of the independent variable on the previous day
(B2
), the between-persons moderator (B3
), the interaction between
the between-persons moderator and the within-person effect (B4),
an error term (v0
) that reflects the deviation of each persons
average from the overall average, an error term (v2
) that reflects
the deviation of each persons slope from the average slope for
Ct1
, and an error term (u) that reflects each persons daily
deviation from his or her own mean on Y. The values of the
3 At the time Barnett et al. (1993, 1995) initially published their work,
multilevel modeling programs such as HLM (Raudenbush & Bryk, 2002)
only allowed for two levels of analysis. Thus, researchers with nested
couple data had to manipulate the program into implicitly allowing a
three-level data structure. They did this by including dummy vectors
representing gender in their Level 1 equations and then using cross-product
terms to separate out the regression equations for men and women. What
this dummy vector approach effectively did was create an additional two
levels at Level 1 and, thus, allow the researchers to estimate two simulta-
neous regressions. Now, however, multilevel modeling programs, includ-
ing HLM and MlwiN, can accommodate many levels of data. Although the
conceptual structure of our design puts couple at the top level, MlwiN
efficiently accommodates simultaneous analyses across gender by treatingmens and womens dependent variables as multivariate outcomes at
Level 1 (and that is what we did in our analysis). Our approach and the
Barnett et al. approach, then, are identical in their multilevel structure
both create simultaneous regression lines for men and women. (To some
extent, the specific description of the nested structure is an arbitrary part of
the terminology: i.e., is gender best thought of as the top level, above
individual, or as the bottom level, nested within each day?) It is important
to note that our approach differs from the Barnett et al. approach, as these
researchers examined growth curves (i.e., how one variable changes over
time). We are interested in whether the relationship between variables
differs across time (i.e., whether one days events predict the next days
feelings for some people but not others).4 We used a logically parallel set of analyses to test whether the intercept
terms were significantly random across couples. The intercepts were sig-
nificantly random in all of the models, and, thus, we assumed randomintercepts as our starting point for presentation.
5 In centering the daily variables around each persons mean, we are
following the recommendations of Barnett et al. (1993, 1995). Centering
the X variables in this way breaks responses at a daily level into two
components: (a) the average level, and (b) changes on each day from the
average level. The dependent variable in any given equation, then, reflects
changes on each day from each persons average level. As a result, this
means of centering the data allows us to examine whether between-persons
differences in mean levels moderate responses at a daily level.6 It is a seven degrees of freedom test because allowing these coeffi-
cients to be random adds two variance terms (the variance in womens and
mens slopes) and five covariance terms to the model.
132 MURRAY, BELLAVIA, ROSE, AND GRIFFIN
-
7/27/2019 Murray Once Hurt, Twice Hurtful
8/22
between-persons variables (ZW
and ZM
) are centered around the
sample mean for women and men, respectively.
Significant effects involving between-persons variables are thus
interpreted as an effect of being high or low on a particular
variable relative to the sample mean.7 Finding a significant inter-
action in the above analysis would mean that the nature of a
cross-day effect, such as the effect of experiencing a higher than
average level of conflict today on tomorrows experience of re-
jection, differed as a function of a between-subjects variable, suchas feeling more or less valued in a partner s eyes than average.
As the last stage of analysis, we tested for gender differences in
the coefficients of interest. We did this by comparing the deviance
of a model that constrained a particular coefficient to be equal for
men and women with the deviance of a model that let this coef-
ficient vary by gender (a one degree of freedom test). The coeffi-
cients for women and men were similar in the large majority of
cases, and, as a result, we typically present coefficients that are
pooled across gender.
The EventFelt Vulnerability Link: Making Mountains out
of Molehills?
We expected perceptions of a partners regard for the self toshape how much (or how little) intimates read into daily negative
but not positive relationship events. Was this the case? Following
the strategy illustrated in Equations 1 and 2, we first tested to see
whether the within-person slopes of interest differed significantly
across individuals.8 That is, we conducted separate analyses pre-
dicting each daily index of felt vulnerability (i.e., felt rejection,
anxiety about acceptance, and felt acceptance) from the prior days
level of felt vulnerability, the prior days events (i.e., conflict, the
partners actual and perceived negative behaviors, the partners
actual and perceived positive behaviors, or the partners negative
mood). (The distinction between the partners actual and perceived
negative and positive behaviors refers to the distinction between
Harrys own description of his own behavior toward Sally and
Sallys perception of Harrys behavior toward her.) These analyses
revealed significant variation in slopes across people for the prior
days conflicts, the partners perceived negative behaviors, and the
partners actual mood on each measure of todays vulnerability
(allowing us to test for the possible moderating effects of per-
ceived regard for these three types of events).9
Accordingly, we then examined whether feeling valued in apartners eyes moderated how vulnerable intimates felt in response
to these events (by inserting the main and interactive effects of
perceived regard into Equations 3 and 4). Table 1 summarizes
these analyses, organizing the equations such that each precipitat-
ing event (i.e., conflict, the partners perceived negative behaviors,
7 In centering womens and mens between-persons variables around
each respective sample mean, we are also following the recommendations
of Barnett et al. (1993, 1995). With this means of centering, the intercept
for women refers to the average level of the dependent variable for women,
and the intercept for men refers to the average level of the dependent
variable for men. The intercepts would not have this easily understood
meaning if we centered around the overall sample mean.8 If every participant completed every diary across the 21 days, we
would have 42 records per couple, for a total of 3,234 records for the
sample. Of course, the number of DERs (and the number of individual
items completed on each record) varied from day to day. As a result, the
number of DERs included in each analysis varies because of missing data
(from 86.4% to 89.1% of all possible records included in the analysis).9 We could not examine whether perceptions of a partners regard
moderated how vulnerable intimates felt in response to the partners actual
(i.e., self-reported) negative behaviors, the partners actual positive behav-
iors, or the partners perceived positive behaviors (because the magnitude
of these cross-day slope coefficients did not vary consistently and signif-
icantly across the sample).
Table 1
Predicting Todays Felt Vulnerability From the Prior Days Events, Perceived Regard, and the Cross-Level Interaction
Todays vulnerability (DV) Intercept (B0)
Prior daysvulnerability
Prior daysevent Perceived regard
Prior DaysEvent
Perceived RegardR2
slopesa
R2
interactionb
B1 SE B2 SE B3 SE B4 SE
IV: ConflictFelt rejection 1.97 .151*** .015 .023 .044 .250*** .041 .090** .035 .037 .075Anxiety over acceptance 1.87 .224*** .014 .006 .029 .304*** .049 .048* .024 .031 .044Felt acceptance 4.73 .227*** .014 .040 .030 .391*** .055 .064* .026 .016 .070
IV: Partners negative behaviorFelt rejection 1.97 .159*** .016 .025 .019 .247*** .041 .028* .014 .028 .093Anxiety over acceptance 1.87 .220*** .014 .013 .011 .304*** .049 .022* .009 .017 .063Felt acceptance 4.73 .231*** .014 .027* .013 .387*** .055 .021 .011 .006 .100
IV: Partners negative moodFelt rejection 1.97 .146*** .013 .042* .018 .262*** .041 .034* .016 .010 .068Anxiety over acceptance 1.87 .208*** .013 .027 .015 .298*** .048 .019 .013 .021 .033
Note. Unless noted otherwise, the coefficients in this table (and all following tables) are pooled across gender. DV dependent variable; IV independent variable.a The R2 for slopes represents the proportional reduction in daily-level variation in the dependent measure associated with specifying the within-person
slopes as random rather than fixed (Snijders & Bosker, 1999).b The R2 for the cross-level interaction represents the proportional reduction in person-level variance across the within-person slopes that is associatedwith perceptions of the partners regard. p .10. * p .05. ** p .01. *** p .001.
133PERCEIVED REGARD AND MARITAL INTERACTIONS
-
7/27/2019 Murray Once Hurt, Twice Hurtful
9/22
and the partners negative mood) predicts each index of situated
felt vulnerability (i.e., felt rejection, anxiety about acceptance, and
felt acceptance). We present the coefficients and standard errors
for all of the terms in each equation in this table, although we only
discuss the expected interactions in detail. (Dividing each coeffi-
cient by its standard error produces the equivalent of a z statistic as
a test of significance). In most cases, the coefficients are pooled
across gender. When we found significant gender differences, we
note the coefficients for men and women in the tables with sub-
scripts (M and W, respectively).10
Reactions to conflict. As we expected, chronically feeling
valued in a partners eyes did indeed moderate how vulnerableintimates felt in response to more conflict-ridden days. That is, the
pooled, cross-level Perceived Regard Conflict interaction term
was significant, predicting felt rejection, anxiety about acceptance,
and felt acceptance (see the first major section in Table 1). Fig-
ure 2a illustrates the nature of this effect for intimates one standard
deviation above and below the mean on perceived regard. As we
expected, detecting a relationship difficulty one day (in this case,
greater levels of conflict than were customarily experienced)
seemed to exacerbate differences between intimates who felt more
or less valued in terms of how much residual hurt they reported on
subsequent days. The results for anxiety and felt acceptance
yielded similar patterns, although the findings for felt acceptance
were a mirror image of these results. (Incidentally, the strong main
effects of feeling positively regarded on traits in these equations
support our conceptualization of this measure as an index of
belongingness: Intimates who believed their partner saw them
more positively on specific traits felt significantly more loved, less
rejected, and less anxious about acceptance across days. The
opposite was the case for intimate who felt less positively
regarded.)
We then decomposed these interactions into their two compo-
nent sets of simple effects by adapting the general procedures
outlined by Aiken and West (1991). We first examined whether
10 We also examined the level of dependence between partners on a
selected subset of the primary intercept and slope coefficients. We found
moderate levels of dependence in both intercept terms and slope terms. The
between-partners correlations in the average level of felt rejection, felt
acceptance, anxiety over acceptance, self-reported negative behavior, and
closeness, for instance, ranged from .25 to .65. The between-partners
correlations in the effects of conflict, perceived rejecting behavior, and
partners negative mood on felt rejection ranged from .17 to .44. The
between-partners correlations in the effects of felt rejection on self-
rejecting behavior and closeness ranged from .08 to .31.
Figure 2. Predicting felt rejection from the prior days threat, perceived regard, and the cross-level interaction.
134 MURRAY, BELLAVIA, ROSE, AND GRIFFIN
-
7/27/2019 Murray Once Hurt, Twice Hurtful
10/22
stressful events have different acute effects for intimates who
chronically feel more or less valued. That is, we examined the
simple effect of yesterdays conflicts on todays feelings of vul-
nerability separately for intimates who chronically felt more val-
ued and those who chronically felt less valued by computing two
conditional regression lines. We did this simply by transforming
the perceived regard scores to represent people one standard de-viation above and below the sample mean for each gender. We
then reconducted the original analysis but substituted the condi-
tional terms for the original terms. In these equations, the coeffi-
cient for conflict is the slope of the regression line linking yester-
days conflict and todays vulnerability, which is conditional on
being either chronically low or high on perceived regard.
The results of these simple effects tests (see the first two
columns of Table 2) illustrate the buffering effects of belonging-
ness: The pooled simple slopes linking yesterdays conflicts to
todays felt vulnerability were significant and in the anticipated
direction for intimates high on perceived regard. That is, intimates
who chronically felt more valued reported significantly greater
confidence in their partners acceptance and significantly less fear
of rejection after high conflict days than after low conflict days.The opposite tendency emerged for intimates who felt less valued,
although the simple slopes for yesterdays conflicts were not
significant.
We then decomposed the same interaction in a different way by
examining whether stressful events accentuate the link between
chronic perceptions of being valued and the acute experience of
vulnerability. That is, we examined the slope of the regression line
linking chronic feelings of being valued to acute vulnerability
separately for days following low versus high threat events (in this
case, on days after intimates reported less vs. more conflict than
they normally experienced). Again, we computed conditional
terms simply by transforming the conflict scores to represent prior
days one standard deviation below and above the sample mean foreach gender. We then reconducted the original analysis using these
conditional terms in place of the original terms. In these condi-
tional equations, the coefficient for perceived regard is the slope of
the regression line linking chronic feelings of being valued to acute
vulnerability that is conditional on the prior day being high or low
on conflict.
The last two columns of Table 2 contain the results of these
simple slopes tests. Decomposing the interaction in this way re-vealed that all of the simple slopes predicting felt vulnerability
from perceived regard were significant for days following low
threat days. That is, intimates who chronically felt more valued felt
less acutely vulnerable following low threat days. However, the
association between feeling chronically valued and feeling acutely
vulnerable was more pronounced for days following high levels of
conflict. As we expected, then, the protective effect of feeling
valued became all the more apparent on days after intimates
reported unusually high levels of conflict and thus seemed to have
evidence at hand to justify anxiety.
Reactions to an ill-mannered partner. What happened when
we substituted perceptions of a partners negative behaviors as the
potential threat to be detected (or discounted)? As the second
major section in Table 1 illustrates, we found significant PerceivedRegard Partners Negative Behavior interactions predicting felt
rejection and anxiety over acceptance, and we found a marginal
interaction predicting felt acceptance. Figure 2b illustrates the
negative behaviorsfelt rejection link for intimates low and high
on perceived regard as an example of the pattern that emerged.
Further illustrating the buffering effects of belongingness, inti-
mates who felt more valued actually reported feeling significantly
more accepted, less rejected, and less anxious about their partners
acceptance on days after they thought their partner was being
particularly ill-behaved, as compared with low-threat days (see
Table 2). That is, the pooled simple slopes for the partners
negative behavior were significant for highs. The opposite ten-
dency again emerged for intimates who felt less valued, althoughthe simple slopes for the partners negative behavior were not
Table 2
Simple Slopes for the Significant Interactions in Table 1
Todays vulnerability (DV)
Prior days event predictingtodays felt vulnerability
Perceived regard predictingtodays felt vulnerability
Lowperceived regard
Highperceived regard
Lowthreat days
Highthreat days
Simpleslope SE
Simpleslope SE
Simpleslope SE
Simpleslope SE
IV: ConflictFelt rejection .071 .055 .118* .060 .215*** .042 .286*** .044Anxiety over acceptance .045 .037 .056 .040 .285*** .049 .323*** .051Felt acceptance .027 .038 .106* .042 .366*** . 055 .416*** . 057
IV: Partners negative behaviorFelt rejection .004 .022 .055* .026 .219*** .042 .274*** .044Anxiety over acceptance .010 .013 .036* .016 .282*** .049 .326*** .051Felt acceptance .004 .014 .049** .019 .366*** .055 .408*** .057
IV: Partners negative moodFelt rejection .077** .023 .006 .026 .237*** .042 .287*** .042
Note. DV dependent variable; IV independent variable. Low and high refer to one standard deviationabove and below the appropriate sample means.* p .05. ** p .01. *** p .001.
135PERCEIVED REGARD AND MARITAL INTERACTIONS
-
7/27/2019 Murray Once Hurt, Twice Hurtful
11/22
significant for lows. Decomposing the interaction into simple
slopes for perceived regard revealed that the link between chronic
perceptions of the partners regard and acute vulnerability was
again exacerbated on days after intimates perceived a particularly
ill-behaved partner and thus had seemingly greater reason to
question their partners acceptance.11
Reactions to a moody partner. Intimates who chronically felt
less valued even read too much into the most seemingly ambigu-
ous event of allthe partner simply feeling badly about himself or
herself on a particular day. As the last section in Table 1 illustrates,
the Perceived RegardMoody Partner interaction was significant
predicting felt rejection. Figure 2c contains illustrative regression
lines. A parallel but not significant interaction emerged, predictingfelt anxiety. (The within-person slopes were not significantly ran-
dom for felt acceptance.) As Table 2 illustrates, intimates who felt
less valued reported significantly greater feelings of hurt and
rejection on days after their partner reported more than his or her
typical amount of negative mood, as compared with low threat
days. That is, the pooled simple slope for the partners negative
mood was significant for lows. The opposite effect tended to
emerge for intimates who felt more valued, although the simple
slope for the partners negative mood was not significant. More-
over, the link between chronic perceptions of the partners regard
and the acute experience of vulnerability was again exacerbated by
the occurrence of a potentially threatening event. That is, the
simple slopes for perceived regard were consistently stronger
following days when the partner had been in a particularly badmood, as compared with low threat days.
Alternative explanations? The results to this point suggest that
the image of themselves people see in a partners eyes shapes how
much (or how little) they read into events that might seem to give
them reason to doubt a spouses affections. As we expected,
intimates who felt more valued found even greater reason to trust
in a partners acceptance on days following threatening events.
The opposite was true for intimates who felt less valued.
There are at least four alternative explanations for these results
that should be examined. The first stems from the fact that we
specified the stability paths in these equations (i.e., the prior days
vulnerability predicting todays vulnerability) as a fixed rather
than a random effect. We took this step to reduce potential mul-
ticollinearity among the predictors, stabilize the iteration process,
and simplify presentation. But taking this measure raises the pos-
sibility that the apparent role of perceived regard in moderating
reactions to relationship threats might really reflect the fact that the
daily experience of felt vulnerability is more stable for highs than
lows (or vice versa). Accordingly, we estimated all of the models
involving significant interactions, setting the stability lag to be
random rather than fixed. The first column of Table 3 contains the
resulting coefficients for the hypothesized cross-level interactions.
The estimation procedure converged in all but one analysis, and we
continued to find significant interactions between perceived regard
and the prior days events predicting felt vulnerability on subse-
quent days.12
The second alternative stems from the possibility that the dy-
namics we observed might be more a function of event frequency
than the chronic activation of belongingness needs. From this
11 Some readers might be concerned by the fact that partner was
irritated/angry with me was included both as one of the events in the
partner negative behavior composite and as one of the feelings in the felt
rejection composite. (Participants responded to two separate items.) After
all, this kind of item overlap might inflate the association between the
composite measures across days. We are not all that troubled by this
possibility. First, we predicted todays felt rejection simultaneously from
felt rejection on the prior day and the level of negative partner behavior
perceived on the prior day. Any cross-day effect of perceived negative
behaviors thus reflects the unique effects of events on subsequent feelings,
with any initial degree of overlap between the events and the feelings
controlled for. Second, we were predicting cross-level interactions, not
simple cross-day main effects. And third, we found similar support for our
hypotheses on measures, such as the link between perceived rejecting
behavior and felt acceptance, that contained no overlapping item content.12 The model predicting anxiety about acceptance from the prior days
level of negative partner behaviors failed to converge to a reliable solution.
Accordingly, we simplified the model and simply examined whether per-
ceived regard moderated the size of the stability lag. It did not.
Table 3
Exploring Alternative Explanations for the Cross-Level Interactions Presented in Table 1
Todays vulnerability (DV)
Stability lagrandom
Controllingevent
frequency
Controllingpartnersappraisal
Controllingglobal
self-esteem
B4 SE B4 SE B4 SE B4 SE
IV: ConflictFelt rejection .088* .033 .092* .036 .088* .036 .067 .038Anxiety over acceptance .047* .024 .055* .025 .048 .025 .030 .026Felt acceptance .048 .025 .065* .027 .061* .027 .050 .028
IV: Partners negative behaviorFelt rejection .034* .013 .027 .015 .027 .015 .028 .016Anxiety over acceptance .021* .010 .022* .010 .019 .010Felt acceptance .016 .010 .020 .011 .019 .011 .021 .012
IV: Partners negative moodFelt rejection .042** .015 .030 .017 .028 .017 .038* .016
Note. Betas are for the Prior Days Event Perceived Regard interaction. DV dependent variable; IV independent variable. p .10. * p .05. ** p .01.
136 MURRAY, BELLAVIA, ROSE, AND GRIFFIN
-
7/27/2019 Murray Once Hurt, Twice Hurtful
12/22
perspective, intimates who feel more valued may rebound more
quickly from threatening relationship events because such events
are more unusual for them, not because they construe those events
more benignly (a variant on regression to the mean). To examine
this possibility, we added each persons (or partners) mean for the
event category and the interaction between mean levels of the
event category and the prior days level of the event to Equations 3and 4. For example, we added terms indexing the partners mean
reports of negative mood each day and the Partners Mean
Mood Prior Days Mood interaction term to the equations
predicting felt rejection. In these (highly conservative) analyses,
feeling more or less valued in a partners eyes still moderated how
hurt intimates felt on days after threatening events. As the second
column of Table 3 illustrates, all of the observed Event Per-
ceived Regard interactions were still significant or at least mar-
ginally significant.
The third alternative stems from the possibility that the effects
of feeling valued might simply be a proxy for the effects of
possessing a more or less valuing partner. If that were the case,
intimates who felt more valued might rebound more readily from
negative events because their partner actually does see them morepositively and thus values them more (and not because feeling
valued on specific traitsand thus deactivating belongingness
needs confers resilience). On the other side of the coin, intimates
who felt less valued might read more into negative events because
their partner actually does see them more negatively and intended
to convey such rejecting attitudes. Consistent with this possibility,
intimates who felt more valued in this sample were, in fact,
regarded more positively by their partner, r(152)W .57,
r(152)M .54, p .001. To examine this alternative, then, we
added the partners actual appraisal of the participant on the
interpersonal qualities scale and the Partners Actual Appraisal
Prior Days Events interaction term to Equations 3 and 4. Despite
the highly conservative nature of these analyses, all of the Per-ceived Regard Event interactions in these analyses were still
significant (or at least marginally significant), as the third column
of Table 3 illustrates, suggesting that the effects of feeling valued
are not masking the effects of actually being valued by ones
partner.
The fourth alternative is that the apparent benefits of feeling
valued in a partners eyes might actually reflect the benefits of
feeling valuable in ones own eyes. After all, low self-esteem
dating and married intimates underestimate how positively their
partner sees them on the specific traits included in our measure of
perceived regard (Murray, Holmes, & Griffin, 2000). To explore
this possibility, we added global self-esteem and the relevant
interaction terms to Equations 3 and 4. The final column of Table 3
contains the Event Perceived Regard interactions for theseanalyses. Despite the fact that intimates with higher global self-
esteem believed their partner saw them more positively on these
interpersonal traits, r(152)W .41, r(152)
M .38, ps .001,
these analyses still revealed consistent, albeit marginally signifi-
cant, Event Perceived Regard interactions. The cross-level
Event Global Self-Esteem interactions were not significant. We
even found a parallel pattern of results when we substituted self-
perceptions on the interpersonal qualities scale for global self-
esteeman even more conservative analysis, given how strongly
self-perceptions on these traits colored impressions of a partners
regard for the self, r(152)W .77, r(152)
M .74. These analyses
suggest that feeling valued in a partners eyes confers a resilience
(or vulnerability) that is not simply isomorphic with self-esteem.
The Felt VulnerabilityCloseness Link: Turning Molehills
Into Mountains?
We expected people who chronically felt less valued to respond
to acute feelings of hurt and vulnerability by rejecting their partner
and thus devaluing what they fear they might lose. In contrast, we
expected people who chronically felt more valued to resist this
temptation to lash out and instead to draw closer in response to
feeling acutely vulnerable. Was this the case?
We first examined whether the within-person effects of interest
differed significantly across couples. That is, we conducted sepa-
rate analyses predicting each daily index of approach and avoid-
ance responses (i.e., own negative behaviors, own positive behav-
iors, anger toward the partner, and feelings of closeness) from the
prior days approach and avoidance responses and the prior days
feeling of vulnerability (i.e., felt rejection, anxiety about accep-
tance, and felt acceptance). All but one of these analyses revealedsignificant random variation across people in the size of the slope
coefficients for felt vulnerability. (The magnitude of the cross-day
slope linking anxiety to positive behaviors did not differ signifi-
cantly across people). We then examined whether feeling valued in
a partners eyes moderated whether intimates responded to feeling
vulnerable by drawing closer to their partner or by pushing their
partner away. Table 4 summarizes these analyses, organizing the
equations such that precipitating feeling predicts each type of
approach or avoidance response. Table 5 presents the correspond-
ing simple effects tests. Again, we focus our discussion on the
expected, significant cross-level interactions.
Own negative behaviors. Did intimates who felt less valued
respond to feelings of vulnerability by engaging in cold, hurtful,and otherwise negative behaviors toward their partner (behaviors
that could, ironically, elicit the very rejection they feared)? As
Table 4 illustrates, the Perceived Regard Yesterdays Felt Re-
jection interaction and the Perceived Regard Yesterdays Anx-
iety interaction were both significant, and the Perceived Regard
Yesterdays Felt Acceptance interaction was marginally signifi-
cant. Figure 3a illustrates the link between felt rejection and
consequent negative behaviors for intimates who chronically felt
more and less valued.
Illustrating the ironic consequences of unfulfilled belongingness
needs, intimates who chronically felt less valued were significantly
more likely to behave in an ill-tempered way toward their partner
on days after they had felt particularly rejected or anxious about
acceptance, as compared with low vulnerability days. That is, thesimple slopes for the prior days felt rejection and anxiety were
significant and positive for people low on perceived regard (see
Column 1 of Table 5). Intimates who felt more valued, however,
were able to resist this impulse. That is, the simple slopes for the
prior days rejection and anxiety were not significant for highs,
suggesting that one days hurts need not translate into the next
days retaliatory actions if people are sufficiently confident of their
partners regard. Put in terms of the simple slopes for perceived
regard, the tendency for intimates who felt less valued to behave in
a more cold, critical, and hurtful way after low vulnerability days
was exacerbated on days after they felt most vulnerable and, thus,
137PERCEIVED REGARD AND MARITAL INTERACTIONS
-
7/27/2019 Murray Once Hurt, Twice Hurtful
13/22
presumably had the greatest need to elicit signs of a partners
approval and affection.
Anger toward the partner. Did intimates who chronically felt
less valued also respond to feelings of vulnerability by finding
greater fault in their partner, effectively blaming them for relation-
ship difficulties? As Table 4 illustrates, the Perceived Regard
Prior Days Felt Rejection interaction was significant for women
but not men. Decomposing the interaction revealed that women
who felt less valued reported significantly greater anger toward
their partner on days after they felt more rejected than usual, as
compared with low vulnerability days. That is, the simple slope for
felt rejection was significant. Decomposing the interaction intosimple slopes for perceived regard, moreover, revealed that the
link between chronic feelings of being valued and anger expres-
sion was exacerbated by the acute experience of heightened hurt
and vulnerability, as we expected.
Feelings of closeness. Did intimates who felt more valued
draw closer to their partner on days after they felt particularly
vulnerable, perhaps as a compensatory defense against feelings of
vulnerability? As we expected, the Perceived Regard Yester-
days Felt Rejection interaction and the Perceived Regard
Yesterdays Anxiety interaction were both significant, predicting
closeness. Figure 3b illustrates the association between the prior
days felt rejection and the next days reports of closeness for
intimates who chronically felt more or less positively regarded. As
the simple slopes in Table 5 illustrate, intimates who chronically
felt more valued actually drew closer to their partner on days after
they felt particularly hurt and sensitized to the possibility of
rejection, as compared with low vulnerability days. That is, the
pooled simple slopes for felt rejection and anxiety were significant
and positive for highs. The opposite tendency emerged for inti-
mates who felt less valued, although the simple slopes for felt
vulnerability were not significant. Put in terms of the simple slopes
for perceived regard, the tendency for intimates who felt more
valued to report greater closeness after low vulnerability days wasexacerbated on days after they had felt most vulnerable to the
possibility of rejection and thus presumably needed the partners
comfort the most.
Own positive behaviors. Although we expected intimates who
felt more valued to also express more caring, affection, and
thoughtfulness toward their partner on days after they felt more
vulnerable, no significant interactions emerged for this variable.
Alternative explanations. Could any of the four alternative
explanations we raised earlier account for the differential re-
sponses to felt vulnerability we observed? When it was possible to
do so, we first estimated random rather than fixed stability paths
Table 4
Predicting Todays Approach and Avoidance Responses From the Prior Days Felt Vulnerability, Perceptions of the Partners
Regard, and the Cross-Level Interaction
Todays response (DV) Intercept (B0)
Prior days DV
Prior days
vulnerability Perceived regard
Prior DaysVulnerability
Perceived
RegardR2
slopesaR2
interactionbB1 SE B2 SE B3 SE B4 SE
IV: Felt rejectionOwn negative behaviors 0.36W .002 .015 .014 .014 .101W*** .024 .034** .012 .027 .130
0.25M .018M .025Anger toward partner 2.0W .013 .020 .110** .032 .227*** .042 .056W* .027 .029 .190W
1.85M .035M .029 .038MCloseness 5.04 .212W*** .020 .012 .018 .204*** .046 .036** .014 .022 .300
.159M*** .020Own positive behavior 1.63 .069*** .014 .021 .018 .145*** .045 .006 .016 .013 .000
IV: Anxiety over acceptanceOwn negative behaviors 0.35W .011 .014 .014 .020 .095W*** .024 .055** .018 .030 .051
0.24M .016M .025Anger toward partner 2.01W .065*** .015 .075 .039 .239*** .041 .037 .031 .025 .030
1.86MCloseness 5.04 .232W*** .020 .053W .034 .210*** .046 .041* .019 .011 .057
.134M*** .020 .045M .031IV: Felt acceptance
Own negative behaviors 0.35W .015W .020 .017 .015 .091W*** .023 .023 .013 .016 .0270.25M .044M* .019 .020M .025
Anger toward partner 2.00W .052*** .015 .090** .029 .224*** .042 .007 .024 .024 .0081.86M
Closeness 5.04 .164W*** .023 .055* .022 .210*** .046 .022 .016 .024 .043.111M*** .023
Own positive behavior 1.63 .066*** .014 .023 .019 .155** .046 .025 .025 .012 .000
Note. Subscripts M and W indicate men and women, respectively. DV dependent variable; IV independent variable.a The R2 for slopes represents the proportional reduction in daily-level variation in the dependent measure associated with specifying the within-personslopes as random rather than fixed (Snijders & Bosker, 1999).b The R2 for the cross-level interaction represents the proportional reduction in person-level variance across the within-person slopes that is associatedwith perceptions of the partners regard.
p .10. * p .05. ** p .01. *** p .001.
138 MURRAY, BELLAVIA, ROSE, AND GRIFFIN
-
7/27/2019 Murray Once Hurt, Twice Hurtful
14/22
for equations in Table 4. As the first column of Table 6 illustrates,
these analyses continued to yield significant Perceived Regard
Prior Days Vulnerability interactions, predicting anger and close-
ness. (The stability paths for participants own negative behavior
were not significantly random).
We then examined whether intimates who felt more valued
responded more constructively to the acute experience of vulner-ability simply because they experienced lower and, thus, possibly
less threatening levels of vulnerability overall. That is, we added
mean levels of felt rejection (and anxiety) and the appropriate
interaction terms to Equations 3 and 4. These highly conservative
analyses still continued to yield significant (or at least marginally
significant) interactions predicting participants own negative be-
haviors and closeness, as the second column of Table 6 illustrates.
We then examined whether intimates who felt less valued re-
sponded more negatively to the acute experience of vulnerability
because their partner actually did evaluate them more negatively.
That is, we added the partners actual appraisal of the intimate and
the interaction terms to Equations 3 and 4. All of the perceived
regard interactions in the third column of Table 6 remained sig-
nificant. We then examined whether intimates who feel less valuedresponded more hurtfully to feeling hurt because they evaluate
themselves more negatively. All of the perceived regard interac-
tions remained significant when we controlled for the main and
interactive effects of global self-esteem (see Table 6, Column 4).
Parallel results emerged when trait-specific esteem on the IQS
served as our measure of self-esteem.
Our focus on behavioral responses to felt vulnerability raises
three further alternative explanations for the results we observed.
The first is that intimates who feel less valued may respond to
vulnerability with hurtful actions because they are defending them-
selves against a battery of real criticisms and complaints (not
because they are overly sensitized to rejection). We conducted a
further set of analyses, in which we added each partners mean
level of self-confessed rejecting behavior and the appropriate in-
teraction terms to Equations 3 and 4 to examine this possibility. As
the fifth column of Table 6 illustrates, we still found significant
perceived interactions in these analyses, suggesting that the retal-
iatory strike that lows take in response to acute vulnerability islargely unwarranted by their partners actual behavior.13
Alternatively, intimates who feel less valued may respond to
acute vulnerability by distancing because they are involved in less
satisfying relationships marked by habitual patterns of recipro-
cated negativity (e.g., Gottman, 1994). After all, our prior search
suggests that intimates who feel less valued report less satisfaction
in their marriage (Murray, Holmes, & Griffin, 2000), and this
dynamic emerged again in the present sample, r(151)W .57,
r(152)M .58. To explore this possibility, we conducted a further
set of analyses, in which we added satisfaction and the Satisfac-
tion Acute Vulnerability interaction to the equations predicting
approach and avoidance responses. The last column of Table 6
contains the coefficients for the perceived regard interactions forthese highly conservative analyses. We still found Felt Vulnera-
bility Perceived Regard interactions predicting reports of inti-
mates negative behaviors toward their partner and their reports of
13 A related possibility is that intimates who feel more valued might
draw closer to their partner on days after they felt more hurt because their
partner is more likely to apologize and engage in reparative behavior when
he or she has behaved badly. This possibility seems unlikely. We did not
find any evidence that people generally behave better on days after they
behaved badly or that the partners of highs behave better on days after
highs felt more hurt.
Table 5
Simple Slopes for the Significant Interactions in Table 4
Todays response (DV)
Prior days felt vulnerabilitypredicting todays response
Perceived regard predictingtodays response
Low
perceived regard
High
perceived regard
Low
threat days
High
threat days
SimpleSlope SE
SimpleSlope SE
SimpleSlope SE
SimpleSlope SE
IV: Felt rejectionOwn negative behaviors .050** .016 .021 .020 .069
W** .025 .134
W*** .028
.010M .026 .046M .028Anger toward partner .149W*** . 042 .016W .054 .192W** .056 .302W*** .059
.093M
* .044 .173M
** .051 .232M
*** .060 .170M
* .067Closeness .026 .021 .050* .025 .172*** .047 .237*** .048
IV: Anxiety over acceptanceOwn negative behaviors .071** .024 .043 .031 .061
W* .027 .129
W*** .026
.018M .025 .050M .029Closeness .010W .079 .097W* .043 .186*** .048 .235*** .048
.088M
** .034 .002M
.039IV: Felt acceptance
Own negative behaviors .041* .018 .008 .021 .073W** .024 .109W*** .026.003
M.027 .036
M.027
Note. Low and high refer to one standard deviation above and below the sample means. Subscripts M and Windicate men and women, respectively. DV dependent variable; IV independent variable. p .10. * p .05. ** p .01. *** p .001.
139PERCEIVED REGARD AND MARITAL INTERACTIONS
-
7/27/2019 Murray Once Hurt, Twice Hurtful
15/22
closeness. These analyses, then, suggest that feeling valued in a
partners eyes and the sense of acceptance or inclusion it reflects
may play a unique role in governing how intimates respond to the
acute experience of vulnerability.
The final possibility is that intimates who feel less valued may
behave in a more cold and rejecting way in response to particularly
hurtful days because hurtful days contain more threatening events.
In other words, the apparent effect of perceived regard in buffering
Figure 3. Predicting approach and avoidance responses from the prior days felt vulnerability, perceivedregard, and the cross-level interaction.
Table 6
Exploring Alternative Explanations for the Cross-Level Interactions Presented in Table 4
Todays response (DV)
Stability lagrandom
Controllingmean
vulnerability
Controllingpartnersappraisal
Controllingglobal
self-esteem
Controllingpartnersbehavior
Controllingsatisfaction
B4
SE B4
SE B4
SE B4
SE B4
SE B4
SE
IV: Felt rejectionOwn negative behaviors .021 .012 .033** .012 .031* .013 .022 .012 .030* .014Anger toward partner .060* .027 .038 .027 .055* .027 .063* .029 .051 .027 .045 .030Closeness .035* .013 .031* .015 .032* .014 .043** .015 .037* .014 .028 .015
IV: Anxiety over acceptanceOwn negative