ArticlePDF Available

Intimacy as an Interpersonal Process: the Importance of Self-Disclosure, Partner Disclosure, and Perceived Partner Responsiveness in Interpersonal Exchanges

Authors:

Abstract and Figures

H. T. Reis and P. Shaver's (1988) interpersonal process model of intimacy suggests that both self-disclosure and partner responsiveness contribute to the experience of intimacy in interactions. Two studies tested this model using an event-contingent diary methodology in which participants provided information immediately after their social interactions over 1 (Study 1) or 2 (Study 2) weeks. For each interaction, participants reported on their self-disclosures, partner disclosures, perceived partner responsiveness, and degree of intimacy experienced in the interaction. Overall, the findings strongly supported the conceptualization of intimacy as a combination of self-disclosure and partner disclosure at the level of individual interactions with partner responsiveness as a partial mediator in this process. Additionally, in Study 2, self-disclosure of emotion emerged as a more important predictor of intimacy than did self-disclosure of facts and information.
Content may be subject to copyright.
Journal of Personality and Social Psychology
1998,
Vol. 74, Mo. 5, 1238-1251
Copyright 1998 by the American Psychological Association, Inc.
0022-35 !4/98/$3.00
Intimacy as an Interpersonal Process: The Importance of Self-Disclosure,
Partner Disclosure, and Perceived Partner Responsiveness
in Interpersonal Exchanges
Jean-Philippe Laurenceau
Pennsylvania State University
Lisa Feldman Barrett
Boston College
Paula R. Pietromonaco
University of Massachusetts at Amherst
H. T. Reis and P. Shaver's (1988) interpersonal process model of intimacy suggests that both
self-
disclosure and partner responsiveness contribute to the experience of intimacy in interactions. Two
studies tested this model using an event-contingent diary methodology in which participants provided
information immediately after their social interactions over 1 (Study I) or 2 (Study 2) weeks. For
each interaction, participants reported on their self-disclosures, partner disclosures, perceived partner
responsiveness, and degree of intimacy experienced in the interaction. Overall, the findings strongly
supported the conceptualization of intimacy as a combination of self-disclosure and partner disclosure
at the level of individual interactions with partner responsiveness as a partial mediator in this process.
Additionally, in Study 2, self-disclosure of emotion emerged as a more important predictor of intimacy
than did self-disclosure of facts and information.
Most theorists and researchers agree that intimacy is an essen-
tial aspect of many interpersonal relationships (e.g., Bartholo-
mew, 1990; Clark & Reis, 1988; McAdams & Constantian,
1983;
Prager, 1995; Reis, 1990; Sullivan, 1953; Waring, 1984).
Nevertheless, considerable variability exists in conceptualiza-
tions of intimacy (Perlman & Fehr, 1987). Some theorists have
defined intimacy as a quality of interactions between persons:
Individuals emit reciprocal behaviors that are designed to main-
tain a comfortable level of closeness (e.g., Argyle & Dean,
1965;
Patterson, 1976, 1982). Other theorists have focused on
Jean-Philippe Laurenceau, Department of Psychology, Pennsylvania
State University; Lisa Feldman Barrett, Department of Psychology, Bos-
ton College; Paula R. Pietromonaco, Department of Psychology, Univer-
sity of Massachusetts at Amherst.
We are grateful to Niall Bolger, Dave Kenny, and Aline Sayer for
their statistical input. We would also like to thank George Levinger for
his comments on a draft of this article.
Data for these studies were collected in part using funding from the
Research and Graduate Studies Office of the College of Liberal Arts,
Pennsylvania State University. Preliminary analyses of the data in Study
1 were presented at the 103rd Annual Convention of the American
Psychological Association, New ^fork, New M)rk, August 1995. Analyses
of a subset of the data in Study 2 were presented at the annual meeting
of the Association for Advancement of Behavior Therapy in Miami
Beach, FL, November 1997.
Correspondence concerning this article should be addressed to either
Jean-Philippe Laurenceau, Department of Psychology, Pennsylvania
State University, University Park, Pennsylvania 16802, or Lisa Feldman
Barrett, Department of Psychology, Boston College, Chestnut Hill, Mas-
sachusetts 02167. Electronic mail may be sent to jxll54@psu.edu or
barretli@bc.edu.
the motivation to seek intimate experiences: People vary consid-
erably in the strength of their need or desire for warm, close,
and validating experiences with other people (e.g., McAdams,
1985;
Sullivan, 1953). Variations also exist in assumptions about
the way in which intimacy develops and is sustained in relation-
ships.
Some theorists propose that intimacy develops primarily
through self-disclosure (e.g., Derlega, Metts, Petronio, & Mar-
gulis,
1993; Jourard, 1971; Perlman & Fehr, 1987), whereas
others have suggested that additional components, such as a
partner's level of responsiveness, contribute significantly to the
development of intimacy in relationships (e.g., Berg, 1987;
Davis,
1982). Furthermore, intimacy has been conceptualized
both as a state or end product of a relationship and as a moment-
to-moment outcome of a process reflecting movement or fluctu-
ation through time (Duck & Sants, 1983).
A recently developed model of intimacy (Reis & Patrick,
1996;
Reis & Shaver, 1988) integrates these multiple perspec-
tives by describing intimacy as the product of a transactional,
interpersonal process in which self-disclosure and partner re-
sponsiveness are key components. In this view, intimacy devel-
ops through a dynamic process whereby an individual discloses
personal information, thoughts, and feelings to a partner; re-
ceives a response from the partner; and interprets that response
as understanding, validating, and caring. Although the Reis and
Shaver (1988) model provides a rich description of how inti-
macy develops on an interaction-by-interaction basis, some of
the hypothesized links have yet to be directly tested in empirical
work. In the present investigation, we test several components
of
the
interpersonal process model of intimacy within the context
of naturally occurring daily social interactions, thus allowing
for an examination of the intimacy process on an interaction-
by-interaction basis.
1238
INTIMACY AS AN INTERPERSONAL PROCESS 1239
The Interpersonal Process Model of Intimacy
According to Reis and Shaver (1988), intimacy results from
a process that is initiated when one person (the speaker) commu-
nicates personally relevant and revealing information to another
person (the listener). The speaker discloses factual information,
thoughts, or feelings and may further communicate emotions
through nonverbal behaviors (e.g., gaze, touch, body orienta-
tion; see Patterson, 1984). As the intimacy process continues,
the listener must respond to the speaker by disclosing personally
relevant information, expressing emotion, and emitting various
behaviors. For the speaker to interpret the listener's communica-
tion as responsive, the listener must convey that he or she under-
stands the content of the speaker's disclosure, accepts or vali-
dates the speaker, and feels positively toward the speaker. At
each stage of this process, personal qualities and individual
differences, including motives, needs, and goals, can influence
each person's behaviors and their interpretation of a partner's
behavior (Reis & Patrick, 1996). Although Reis and Shaver
largely focused on what occurs in any given interaction, they
explicitly acknowledge that intimacy accrues across repeated
interactions over time. As individuals interpret and assimilate
their experiences in these interactions, they form general percep-
tions that reflect the degree to which the relationship is intimate
and meaningful (Reis, 1994). Although these generalized per-
ceptions of intimacy in the relationship develop over the course
of repeated interactions, overtime, they may take on an emergent
property that goes beyond experiences in each individual inter-
action (Chelune, Robison, & Kommor, 1984).
The model emphasizes two fundamental components of inti-
macy: self-disclosure and partner responsiveness (Reis & Pat-
rick,
1996;
Reis & Shaver, 1988). Many definitions suggest that
intimacy is a feeling of closeness that develops from personal
disclosures between people (Perlman & Fehr, 1987). Self-dis-
closure refers to the verbal communication of personally relevant
information, thoughts, and feelings to another; research in this
area has often relied on degree or depth of self-disclosure as
an index of intimacy (Altman & Taylor, 1973; Derlega et al.,
1993;
Jourard, 1971). Some theorists (Morton, 1978; Reis &
Patrick, 1996; Reis & Shaver, 1988) have suggested that particu-
lar types of self-disclosure (i.e., those revealing the core self)
are more closely linked to the experience of intimacy than others.
Researchers have distinguished between factual (i.e., descrip-
tive) and emotional (i.e., evaluative) disclosure when examining
the impact of disclosing the self in intimate relationships (Mor-
ton, 1978; Reis & Shaver, 1988). Factual self-disclosures are
those that reveal personal facts and information (e.g., "I've had
three romantic partners in my life''). Emotional self-disclosures
are those that reveal one's private feelings, opinions, and judg-
ments (e.g., "The last breakup was so painful that I'm not
sure if I can love someone again
1
'). Although both types of
disclosures reveal private aspects of the self to others, disclo-
sures involving emotions and feelings lie most closely at the
core of one's self-definition (Greenberg & Safran, 1987; Reis &
Patrick, 1996). Self-disclosures that involve emotions are be-
lieved to generate greater intimacy than those that are merely
factual because such disclosures open the way for the listener
to support and confirm core aspects of the discloser's view of
self (Reis & Shaver, 1988; Sullivan, 1953).
Partner responsiveness is the other key component in the de-
velopment of intimacy (Berg, 1987; Berg & Archer, 1982;
Davis,
1982; Kelley et al., 1983). Partners are responsive when
their behaviors (e.g., disclosures, expressions of emotion) ad-
dress the communications, needs, wishes, or actions of the per-
son with whom they are interacting (Miller & Berg, 1984).
According to Reis and Shaver (1988), speakers are more likely
to experience an interaction as intimate if they perceive their
partner's response as understanding (i.e., accurately capturing
the speaker's needs, feelings, and situation), validating (i.e.,
confirming that the speaker is an accepted and valued individ-
ual),
and caring (i.e., showing affection and concern for the
speaker). Reis and Shaver regard the speaker's interpretation of
the listener's communication as more important for the develop-
ment of intimacy than a speaker's disclosure or the listener's
actual response. Although a partner may make a genuine attempt
to be responsive to a disclosure, the speaker may not perceive
the partner's behavior as responsive to his or her needs. This
reasoning suggests that the degree to which a listener's actual
communication (what we will call partner disclosure) is associ-
ated with intimacy in the interaction should depend heavily on
the nature of the speaker's perceptions of and feelings about the
partner's response (what we will call perceived partner respon-
siveness). Thus, the extent to which the speaker perceives the
partner as responsive should mediate the association between
the partner's disclosure and level of intimacy in the interaction.
Despite the conceptual appeal of Reis and Shaver's (1988)
interpersonal process model of intimacy, little empirical research
has addressed its validity. One unpublished investigation of the
model (Lin, 1992), however, suggests that both self-disclosure
and partner responsiveness contribute to perceptions of intimacy
at the level of the relationship. In this study, college students
reported on their level of self-disclosure and perceptions of their
partner's responsiveness after every social interaction over a 10-
day period and, in a later interview, indicated the degree of
relationship intimacy with each interaction partner. Both
self-
disclosure and perceived partner responsiveness, aggregated
across daily interactions, predicted overall relationship inti-
macy: People who, on average, disclosed more and perceived
a partner to be responsive reported greater intimacy in their
relationship with that partner. In addition, there was some evi-
dence that emotional disclosures were more important to rela-
tionship intimacy than were disclosures about facts. Although
this study provided initial evidence that both self-disclosure and
partner responsiveness are central components of intimacy at
the level of the relationship, other important aspects of the model
remain to be validated. In the present work, we sought to test
the Reis and Shaver model by examining the links among several
components of the intimacy process (i.e., self-disclosure, part-
ner disclosure, partner responsiveness) on an interaction-by-
interaction basis. It is particularly important to investigate these
links at the level of interactions to determine whether feelings
of intimacy in a given interaction are associated with the percep-
tion that self-disclosure has occurred and the perception that
the partner returned the disclosure and was responsive.
1240
LAURENCEAU, FELDMAN BARRETT, AND PIETROMONACO
Overview of Studies 1 and 2
The overarching goal of the present studies was to test Reis
and Shaver's (1988) interpersonal process model of intimacy
at the level of individual social interactions. In Study 1, we
examined two key tenets of the Reis and Shaver model: (a)
Self-disclosure and partner disclosure will contribute to feelings
of intimacy on an interaction-by-interaction basis, and (b) this
process will be mediated by the degree to which a partner is
perceived as responsive. In Study 1, partner responsiveness was
operationalized solely by the degree to which a partner was
perceived as accepting. In Study 2, we replicated and extended
the findings from Study 1 by relying on broader measures of
perceived partner responsiveness and self-disclosure and by in-
vestigating a third hypothesis: (c) Emotional disclosures will
contribute more to intimacy in the interaction than would factual
disclosures.
We examined these aspects of the model by having partici-
pants complete a version of the Rochester Interaction Record
(RIR; Reis & Wheeler, 1991). The RIR is an event-contingent
diary that participants complete immediately following their
social interactions over a specified period of
time.
In our studies,
after every interaction, participants provided detailed informa-
tion about their degree of self-disclosure, their perceptions of
partner's disclosure, the degree to which they perceived their
partners as responsive toward them, and their feelings of inti-
macy over a 1 (Study 1) or 2 (Study 2) week period. The RIR
methodology allowed us to examine the hypothesized relation-
ships among self-disclosure, partner disclosure, partner respon-
siveness, and intimacy on an interaction-by-interaction basis.
Study 1
In Study 1, we tested two basic hypotheses following from
the Reis and Shaver (1988) model. First, we tested the hypothe-
sis that both self-disclosure and partner disclosure predict feel-
ings of intimacy. Figure 1A shows the path model representing
the hypothesized links among self-disclosure, partner disclosure,
and intimacy. The degree to which individuals self-disclose
should contribute to their feelings of intimacy, after controlling
for their perceptions of the partner's disclosure (path p^). The
degree to which individuals view the partner's disclosing in turn
also should contribute to feelings of intimacy, after controlling
for self-disclosure (path p
32
).
Second, we tested the hypothesis that individuals will experi-
ence intimacy in an interaction when self-disclosure and partner
disclosure are linked to feelings that the partner is being respon-
sive.
Reis and Shaver (1988) emphasized the potential mediating
role of partner responsiveness in the relationship between part-
ner disclosure and intimacy. In addition, it is possible that part-
ner responsiveness might also mediate the link between
self-
disclosure and intimacy, and for completeness, we tested this
possibility. The mediational model, shown in Figure IB, in-
cludes paths representing (a) the effect of self-disclosure (p^)
and partner disclosure (pn) on perceived partner responsiveness
and (b) the effect of each variable (i.e., self-disclosure, partner
disclosure, and perceived partner responsiveness) on intimacy,
while controlling for the effect of the others (paths p
AX
, p
42
, and
p
43
).
For a partial mediation effect, perceived partner respon-
siveness should be related significantly to ratings of self-disclo-
sure and to partner disclosure (paths p
3
i and p
32
), as well as to
ratings of intimacy (path/>
43
; Kenny, Kashy, & Bolger, 1997).
For a full mediation effect (Baron & Kenny, 1986; Kenny et
al.,
1997), there is an additional requirement that self-disclosure,
partner disclosure, or both no longer have a significant direct
effect on intimacy (i.e., path
p
4lt
p
42
, or both should not be
significantly different from zero).
Method
Participants
The study began with 104 participants who were selected from a
larger undergraduate participant pool, 56 sampled from the University of
Massachusetts and 48 sampled from the Pennsylvania State University.
1
Fourteen percent of the sample (15 participants) did not complete the
study. These participants did not differ from those who remained in the
study on any variable relevant to the present investigation. Twenty-one
percent of the remaining sample (19 participants) reported having used
their memory to complete more than 25% of the interaction records. We
removed these participants from the analysis to minimize the influence
of recall bias on participants' reports. The final sample consisted of 69
participants (42 women) who had complete data for the interaction
record ratings on their dyadic interactions. All participants received extra
credit for their participation and had a chance of winning $50 in a lottery
at the end of the semester.
Interaction Record
We adapted the RIR to assess self-disclosure, partner disclosure, feel-
ing accepted by a partner, and intimacy experienced during social inter-
actions. Participants completed the fixed-format interaction record after
every interaction lasting 10 min or longer (Reis & Wheeler, 1991) for
a
1-week
(i.e., 7-day) period. We defined an interaction as any encounter
with another person(s) in which the individuals attended to one another
and possibly adjusted their behavior in response to one another (Reis &
Wheelei; 1991). We called the interactions "social" because they in-
volved at least one other person, but the interactions included more than
situations in which the participants socialized for entertainment purposes
(e.g., we sampled interactions at work, over the telephone, during
classes, on errands). Using the RIR, participants provided information
about the number of interaction partners present, the initials of partners
for each interaction, and who initiated the interaction. We analyzed only
dyadic interactions, because theories of intimacy focus primarily on
dyadic exchanges. Fifty-eight percent of the interactions (n =
1,494)
from the total data set were used for the present report.
For each interaction, participants rated a variety of interaction aspects
on 5-point scales (1 = very little, 5 = a great deal). Only the RIR
items of interest for this report are presented here (see Appendix A for
exact wording of Study 1 RIR items).
1
These data were originally collected to study adult attachment styles.
The sample includes approximately equal numbers of individuals from
each of four attachment styles (i.e., secure, preoccupied, fearful-avoid-
ant, and dismissing-avoidant; Bartholomew & Horowitz, 1991). Analy-
ses using the Study I data set appear in two previously published reports
(Feldman Barrett & Pietromonaco, 1997; Pietromonaco & Feldman Bar-
rett, 1997), but the hypotheses and analyses reported in those articles
do not overlap with those presented here. Only those materials relevant
to the present investigation are described here.
INTIMACY AS AN INTERPERSONAL PROCESS
1241
Self-Disclosure
(1)
721
Partner Disclosure
(2)
Self-Disclosure
(1)
Perceived Partner
Responsiveness
JJ43
Partner Disclosure
(2)
Intimacy
(4)
Figure 1. Models containing the disclosure components of the intimacy process (A) and perceived partner
responsiveness as a mediator of the intimacy process (B).
Self-disclosure. Participants rated the amount they disclosed in gen-
eral (one item) and how much they expressed their emotions (one item)
to their partner in the interaction. A summary variable was created using
the average of these two items.
Partner disclosure. Participants rated the amount the interaction
partner disclosed (one item), the amount of positive emotion their partner
expressed (one item), and the amount of negative emotion their partner
expressed (one item). A summary variable was created using the average
of these three items. We combined ratings of partner disclosure and
partner's emotional expression because we were interested in investigat-
ing partner disclosure at a global level.
Perceived partner responsiveness. Participants rated the degree to
which they felt accepted by their interaction partner during the interac-
tion. The item was used as an index of Reis and Shaver's (1988) concep-
tualization of partner responsiveness in this study.
Intimacy. Participants rated the amount of intimacy that they experi-
enced during the interaction.
Procedure
Participants attended three laboratory sessions. During the first ses-
sion, the experimenter explained that the study concerned how people
think and feel about their social interactions with others and that partici-
pants would keep records of all of their social interactions for 7 days.
1242
LAURENCEAU, FELDMAN BARRETT,
AND
PIETROMONACO
Tb preserve confidentiality, participants selected
a
code name
to
write
on
all of
their study materials. Participants also completed several ques-
tionnaires during the first session
(for a
complete description, see Pietro-
monaco
&
Rsldman Barren, 1997). Afterward,
the
experimenter
ex-
plained
the
procedure
for
completing
the
interaction records
and
care-
fully defined
all
items
on the
interaction record form. For example,
the
experimenter indicated that
the
term intimacy referred
to the
extent
to
which the participants felt interpersonally close
to
their interaction part-
ners
in a
given interaction,
and
that intimacy
did not
necessarily refer
to sexual activity.
The
experimenter emphasized
the
importance
of an-
swering honestly when using
the
interaction records
and of
completing
a record
as
soon
as
possible (within
15 min)
after each interaction.
In
addition
to
oral instructions, participants received written instructions
to which they could refer during the course
of
the study. Prior
to
leaving
the first laboratory session, participants received several interaction
re-
cords with which
to
practice.
During
the
second laboratory session, participants reviewed their
practice interaction records with
the
experimenter.
The
experimenter
answered
any
questions
and
gave participants
a
final written
set of
instructions
for
completing
7
days
of
interaction records. Participants
returned their interaction records three times during their recording
week,
and
they received extra lottery tickets
for
returning their forms
on time.
The
experimenter phoned, within
24 hr, any
participants
who
did
not
return their forms
on
time
and
reminded them
to
return
the
forms.
During the third laboratory session, the experimenter interviewed par-
ticipants about their reactions
to the
study. Participants estimated
how
difficult they found
the
study,
how
accurate their recording
was, and
how much their social patterns changed
as a
result
of
being
in the
study.
To ensure that participants followed
all
instructions,
the
experimenter
asked several specific questions about
the
accuracy with which partici-
pants
had
recorded their interactions, including
(a)
whether they
had
recorded
all of
their interactions
and, if
they
had not,
what percentage
they
had not
recorded (percentage
not
recorded:
M =
15.4%,
SD =
14.8)
and (b)
whether they
had
completed any interaction records from
memory (i.e., more than
1 hr
later)
and, if
they
had, the
percentage
of
interaction forms that they
had
completed from memory.
2
The
experi-
menter stressed that participants would
not be
penalized
in any way
(i.e.,
they would still receive credit
and
lottery tickets)
if
they
had not
followed instructions
and
that
we
were simply interested
in
obtaining
an accurate picture
of
their data.
Results
Data Analytic Strategy
The interaction data in this study conformed to a multilevel
data structure (Goldstein, 1987; Kenny et al., 1997). A defining
feature of multilevel data is the existence of a hierarchy of
observations in which multiple lower level observations are
grouped within upper level units. In this research, the lower
level data consisted of participants' ratings of self-disclosure,
partner disclosure, partner responsiveness, and intimacy experi-
enced during social interactions. Each lower level variable was
measured on an interaction-by-interaction basis and, therefore,
consisted of multiple data points for each individual. These
lower level data were nested within upper level units, or partici-
pants. Hierarchical linear modeling (HLM; Bryk & Rauden-
bush, 1987, 1992; Bryk, Raudenbush, & Congdon, 1996) is a
statistical program designed for use with multilevel data sets.
HLM was used to analyze the interaction data because it allowed
us to analyze within-subject (lower level) and between-subject
Table 1
Average Within-Subject Correlations for Items Used
in Composite Variables (Study 1)
Variable
1
1.
Self-disclosure
2.
Self expression
of
emotion
.63**
3.
Partner disclosure
.61**
.43**
4.
Partner expression
of
positive
emotion
.29**
.38**
.31**
5.
Partner expression
of
negative emotion
.01 .06 .09*
-.40**
Note. Variables
1 and 2
constitute
the
self-disclosure composite. Vari-
ables
3, 4, and 5
constitute
the
partner disclosure composite.
*p
<
.05. **/>
<
.001.
(upper level) variation simultaneously, thus enabling us to model
each source of variation while taking the statistical characteris-
tics of the other level into account. HLM first computes a regres-
sion equation for each participant in which a lower level out-
come variable is regressed on lower level predictors; the individ-
ual regression parameters for these predictors are then used to
estimate the average parameter estimates for all participants as
well as the amount of individual variation around this average.
Standardized HLM regression coefficients were used to estimate
paths.
3
Self-Disclosure
and
Partner Disclosure
in the
Intimacy
Process
Using HLM, we calculated the average within-subject corre-
lations for items within the self-disclosure and partner disclosure
composite variables. These correlations are presented in Ta-
ble 1.
To test the hypothesis that both self-disclosure and partner
disclosure emerge as significant predictors of intimacy in dyadic
social interactions, we estimated the paths presented in Figure
1A using several HLM analyses. The within-subject level of the
data analytic model estimated the p
3i
and p
i2
paths using the
formula:
lij
= b
Q
b
2
X
r
ih
(1)
2
Six participants reported that they had recorded
all
interactions (9%
of final sample), 51 participants reported they
had
missed recording
up
to 25%
of
their interactions (73%
of the
final sample),
6
participants
reported that they
had
missed recording between 25%
and
30%
of
their
interactions
(9% of
final sample),
5
participants reported that they
had
missed between 30% and 50% of their interactions (7%
of
final sample),
and 1 participant reported having missed 75%
of
his
or
her interactions.
Thus,
the
majority
of
participants (81%
of
final sample) reported that
they
had
documented three quarters
of
their social interactions over
the
observation week.
3
Because
all RIR
ratings were based
on the
same metric,
we
opted
to submit
the raw
data
to HLM for
analysis
to
obtain unstandardized
estimates and then standardized these estimates using estimates of pooled
within-subject standard deviations
for
the relevant variables.
INTIMACY
AS AN
INTERPERSONAL PROCESS
1243
Table
2
Average Within-Subject Correlations
of
Rochester
Interaction Record Variables
for
Study
1
Variable
1
1.
Self-disclosure
2.
Partner disclosure
3.
Partner responsiveness
4.
Intimacy
.52
.21
.66
.26
.57 .28
Note.
All
correlations significant
atp < .001.
where
/^ is
participant
j's
intimacy rating
on the ith
social
interaction,
b
0
is
participant
j's
average intimacy rating across
all dyadic social interactions, SD
tJ
is
participant
j's
degree
of
self-disclosure
on the ith
social interaction,
b
x
represents
the
relationship between self-disclosure
and
intimacy controlling
for partner disclosure
for
participant,/, PD
Lj
is
participant
j's
perception
of his or her
partner's disclosure
on the ith
social
interaction,
b
2
represents
the
relationship between partner
dis-
closure
and
intimacy controlling
for
self-disclosure
for
partici-
pant
j,
and r
0
is a
within-subject error component.
The between-subjects (upper) level
of the
model allowed
us
to assess
the
average relationship between self-disclosure
and
intimacy
(the
average value of
p^),
and the
average relationship
between partner disclosure
and
intimacy
(the
average value
of
Pn),
as
follows:
~
bio +
U
2
j,
(2)
(3)
where
the
upper level estimate,
&
i0
,
represents
the
average rela-
tionship between self-disclosure
and
intimacy, controlling
for
partner disclosure,
and the
upper level estimate,
b^,
represents
the average relationship between partner disclosure
and
inti-
macy, controlling
for
self-disclosure;
w
y
and
u%
are
between-
subjects error terms
and
represent
the
degree
to
which
the
Level
1 regression coefficients
for
participants varied around
the
aver-
age
for
each coefficient. Although
we
also modeled
the
average
of
the
intercept
{b
m
), we do not
focus
on
this aspect
of the
analysis because
it is not
important
to our
hypotheses. Similar
HLM analyses were used
to
estimate
the
average zero-order
correlations between variables across
all
individuals found
in
Table
2.
These correlations represent
the
average simple rela-
tionship among
the
lower level variables.
The results, presented
in
Figure 2A, indicated that
on
average,
self-disclosure
and
partner disclosure were significantly corre-
lated
{r = .52, p <
.001). Both self-disclosure
and
partner
disclosure were associated significantly with ratings
of
intimacy.
Self-disclosure
was
positively correlated with intimacy
on
aver-
age
(r =
.66,
p < .001) and
positively predicted intimacy over
and above
the
effect
of
partner disclosure,
p
zl
= .49, t =
15.32,
p
<
.001. Similarly, partner disclosure
was
positively correlated
with intimacy
on
average
(r = .57, p < ,001) and
positively
predicted intimacy over
and
above
the
effect
of
self-disclosure,
p
32
=
.28,*
=
9.34,
p < .001.
The Role
of
Perceived Partner Responsiveness
in the
Intimacy Process
To test whether partner responsiveness mediated
the
intimacy
process,
we
estimated
the
paths represented
in
Figure
IB
using
HLM analyses similar
to
those presented
in
Equations
1, 2, and
3.
The results, presented
in
Figure 2B, indicated that
on
average,
perceived partner responsiveness partially mediated
the
relation-
ships among self-disclosure, partner disclosure,
and
feelings
of
intimacy. Self-disclosure
and
partner disclosure were signifi-
cantly correlated with perceived partner responsiveness
(r =
.21,
p < .001, and r = .26, p < .001,
respectively);
self-
disclosure
and
partner disclosure were also uniquely related
to
perceived partner responsiveness,
p
3i
~ .08, t =
2.65,
p < .01,
andp
32
=
-20,
t
431, p <
.001,
respectively. Perceived partner
responsiveness
was
also significantly correlated with intimacy
(r
=
.28,
p < .001) and
demonstrated
a
unique relationship
to
intimacy,
p
43
=
.11,
t =
3.80,
p < .001.
Full mediation
of a
predictor's effect
on the
criterion variable
is indicated
if the
predictor
no
longer
has a
direct effect
on the
criterion after
the
mediator
has
been introduced.
For
example,
if perceived partner responsiveness fully mediated
the
effect
of self-disclosure
on
intimacy, then
the
direct path from
self-
disclosure
to
intimacy
in
Figure
IB
would
be
reduced
to
zero.
Partial mediation exists when
the
direct effect
is
reduced
in
magnitude
yet
remains different from zero. Self-disclosure
and
partner disclosure continued
to
have significant direct effects
on ratings
of
intimacy after
the
contribution
of
perceived partner
responsiveness
was
controlled, p
41
.48, / =
15.49,/?
< .001,
and
p
42
= .26, t = 8.64, p < .001,
respectively, suggesting
that
the
relationship between disclosures
and
intimacy
was
only
partially mediated
by
perceptions
of
partner responsiveness.
To examine specifically
the
significance
of
the partial media-
tion effects,
we
tested whether
the
mediator (i.e., perceived
partner responsiveness) significantly reduced
the
size
of the
direct effects
on
intimacy (Baron
&
Kenny, 1986; Kenny
et al.,
1997).
Tests were conducted
on
partial regression coefficients
using unstandardized HLM estimates with their associated stan-
dard errors
(N.
Bolger, personal communication, September
1997).
To
test whether perceived partner responsiveness medi-
ated
the
relationship between self-disclosure
and
intimacy,
we
compared
the
standardized estimate
of
self-disclosure's effect
on intimacy
in the
original model
(.49)
with
the
corresponding
effect
in the
mediated model (.47). This difference represented
a small
but
significant reduction
(z
2.13,
p <
.02), suggesting
that perceived partner responsiveness partially mediated
the ef-
fect
of
self-disclosure
on
intimacy, after controlling
for the ef-
fects
of
partner disclosure.
The
same test
was
conducted with
partner disclosure
as the
predictor variable.
The
reduction
in
the standardized path
of
partner disclosure
on
intimacy (from
.28
to .26) was
also small
but
significant
(z =
2.82,
p <
.003),
indicating that perceived partner responsiveness also partially
mediated
the
effect
of
partner disclosure
on
intimacy.
Discussion
The findings from Study
1
supported
the
basic tenets
of the
interpersonal model
of
intimacy
at the
level
of
social interac-
1244
LAURENCEAU, FELDMAN BARRETT, AND PIETROMONACO
Self-Disclosure
(1)
.52
VJ
Partner Disclosure
(2)
n
Self-Disclosure
(1)
.47 (.55)
.52
Partner Disclosure
(2)
Perceived Partner
Responsiveness
Intimacy
(.38)
Figure 2. The relationship of disclosure to feelings of intimacy (A) and the estimation of perceived partner
responsiveness as a mediator in the intimacy process (
B),
Study 1. Unstandardized coefficients are presented
in parentheses. All paths were statistically significant.
tions.
Both self-disclosure and partner disclosure significantly
predicted intimacy across a range of interpersonal interactions
and social relationships. The observed mediation effect was not
as large in magnitude as we expected, however. According to
Reis and Shaver (1988), the intimacy process should also de-
pend on a person feeling understood and cared for, in addition
to feeling accepted. It is possible that feeling accepted by a
partner is a complex appraisal that is associated with feeling
understood and cared for. One explanation for the weak mediator
effect, then, is that our measure of partner responsiveness in
Study 1 (i.e., acceptance) was too narrow; a conceptually
broader measure of partner responsiveness might produce a
stronger mediation effect.
Study 2
The goals of Study 2 were to replicate and extend the findings
from Study 1 in several ways. Overall, we used the same basic
path models to evaluate the interpersonal process model of inti-
macy, but with three modifications. First, we increased the diary
TNTIMACY AS AN INTERPERSONAL PROCESS
1245
collection period from I to 2 weeks. Second, we measured
partner responsiveness more broadly and in a manner that more
closely paralleled Reis and Shaver's (1988) conceptualization,
by asking participants about the degree to which they felt under-
stood, cared for, and accepted by their partners during social
interactions. Third, we measured self-disclosure with items that
distinguished descriptive self-disclosures (disclosures of facts)
from evaluative self-disclosures (disclosures of feelings) to
allow for a finer grained analysis of the relative contributions
of different types of self-disclosure. This refined measure al-
lowed us to follow up on previous work (i.e., Lin, 1992; Morton,
1978) that has suggested that emotional self-disclosure would
be more strongly related to intimacy than would factual
self-
disclosure.
Table 3
Average Within-Subject Correlations for Items Used
in Composite Variables (Study 2)
Method
Participants
The study began with 244 participants who were selected from a
larger undergraduate subject pool, 158 sampled from the University of
Massachusetts and 86 sampled from the Pennsylvania State University.
4
Twenty percent of the sample (48 participants) did not complete the
study. Thirty-nine percent of the remaining sample (77 participants)
reported having used their memory to complete more than 25% of the
interaction records. This percentage is somewhat larger than that ob-
served in Study 1, possibly because participants were asked to sample
their social interactions for 2 weeks rather than for 1 week. We removed
these participants' data from the analysis to minimize the influence of
recall bias on participants' reports. Twenty-five percent (30 participants)
of the remaining sample (119 participants) reported no dyadic relation-
ships;
that is, all of the interactions that they recorded involved more
than one person. We suspect that many participants who appeared to
have had no dyadic interactions actually made recording errors in re-
porting the number of interactions partners present (i.e., counting both
themselves and the interaction partner rather than the interaction partner
alone).
Given that it was not possible to determine when such errors
were made, we used a conservative approach and removed all such
individuals' data from the analyses. The final sample consisted of 89
participants (51 women, 38 men) who reported a total of 3,955 dyadic
interactions (68% of all interactions recorded). All participants received
extra credit for their participation and had a chance to win $50 in a
lottery at the end of the semester.
Interaction Record
As in Study 1, participants completed a version of the RIR after each
social interaction lasting 10 min or longer and rated their interactions
using 5-point scales (1 = very little, 5 = a great deal). Only the RIR
items of interest for this report are presented here (see Appendix B for
exact wording of Study 2 RIR items).
Self-disclosure. Participants rated the degree to which they disclosed
facts (one item), how much they expressed their thoughts (one item),
and how much they expressed their emotions (one item) to their partner
in the interaction. An overall summary variable for self-disclosure was
created using the average of these three items. Descriptive self-disclosure
was operationalized using the disclosure-of-facts item; evaluative
self-
disclosure was operationalized using the expression-of-emotions item.
Partner disclosure. Participants rated the degree to which the inter-
action partner disclosed thoughts and emotions (one item).
Perceived partner responsiveness. Participants rated the degree to
which they felt accepted by their interaction partner (one item), how
Variable
1.
Self-disclosure of facts
2.
Self-disclosure of thoughts
3.
Self-disclosure of
emotions
4.
Understanding by partner
5.
Acceptance by partner
6. Cared for by partner
1
_
.20
.10
.14
.08
.02
a
2
.70
.33
.24
.33
3
.29
.22
.40
4
.81
.46
5
.42
6
Note. All correlations are significant at p < .005, except as noted.
Variables I, 2, and 3 constitute the self-disclosure composite. Variables
4,
5, and 6 constitute the partner responsiveness composite.
much they felt understood by their interaction partner (one item), and
how much they felt cared for by their interaction partner (one item)
during each social interaction. A summary variable was created using
the average of these three items.
Intimacy. Participants rated the amount of closeness that they experi-
enced in the interaction. We chose the term closeness rather than intimacy
to ensure that participants understood that we were referring to psycho-
logical proximity and not to sexual intimacy.
Using HLM, we calculated the average within-subject correlations
for items within all composite variables. These correlations are presented
in Table 3.
Procedure
The procedures for data collection in Study 2 were identical to those
of Study 1, with the exception that participants kept records of their
social interactions for a 2-week period rather than a
1-week
period.
5
4
As in Study 1, the participants of Study 2 were not a randomly
selected sample from the student participant pool. This sample was
selected for the primary purpose of obtaining equal numbers of romantic
attachment styles (27% secure, 24% fearful-avoidant, 26% preoccu-
pied, and 24% dismissing-avoidant). To address this potential threat to
the external validity of our findings, we reanalyzed the data using a
random sample of participants that reflected the proportion of romantic
attachment style membership consistently found in the larger undergradu-
ate participant pool: 48% secure, 28% fearful-avoidant, 16% preoccu-
pied, 8% dismissing-avoidant. The results were essentially identical to
those that we report in the main body of the article. In addition, we
inspected the distributions of the key variables and found no evidence
of substantial deviation from normality.
5
The extension of the diary recording period (from 1 to 2 weeks)
increased the number of social interactions available for reporting. This
increase in the number of observations per participant resulted in an
increase in the reliability of the slope estimates for self-disclosure, part-
ner disclosure, and partner responsiveness:
.31,
.27, and .35, respectively,
in Study 1 to .53, .65, and .56, respectively, in Study 2. The reliability
of these change parameters reflects the proportion of observed parameter
variance that is accounted for by true parameter variance. This notion
of reliability differs from that of coefficient, an index that is not appro-
priate for use with within-subject data of this type. The reliability of
slope estimates depends on the number of observations within each
participant and the variance of the predictor variable within each
participant.
1246
LAURENCEAU, FELDMAN BARRETT,
AND
PIETROMONACO
Results
As
in
Study
1, all
analyses were conducted using HLM. Stan-
dardized regression coefficients were used
to
estimate paths.
Table
4
contains
the
average correlations across individuals
for
the
RIR
variables
in
Study
2.
Self-Disclosure
and
Partner Disclosure
in the
Intimacy
Process
First,
we
estimated
the
path relationships among self-disclo-
sure,
partner disclosure,
and
feelings
of
intimacy,
as
presented
in Figure
1A. The
results, presented
in
Figure
3A,
indicated
that
on
average, both self-disclosure and partner disclosure dem-
onstrated significant predictive effects
on
ratings
of
intimacy
across
a
range
of
social interactions. Self-disclosure
and
partner
disclosure were significantly correlated with each other
on
aver-
age
O =
.60,
p < .001) and
were significantly correlated with
intimacy
(r =
.55,
p < .001, and r = .57, p <
.001, respec-
tively).
Intimacy was uniquely predicted
by
both self-disclosure,
p
3i
= .31, t =
10.71,
p <
.001,
and
partner disclosure,
p
32
=
.40,
t =
13.27,
p <
.001.
The
findings, thus
far,
replicate those
found
in
Study
1.
The Role
of
Partner Responsiveness
in the
Intimacy
Process
Next,
we
examined whether perceived partner responsiveness
acted
as a
mediator
of the
intimacy process,
as
portrayed
in
Figure
IB, The
results, presented
in
Figure 3B, replicated those
for Study
1.
Self-disclosure
was
significantly correlated with
perceived partner responsiveness
(r =
.41,
p < .001) and dem-
onstrated
a
unique predictive relationship,