ArticlePDF Available

Couples' shared participation in novel and arousing activities and experienced relationship quality

Authors:

Abstract

Using a newspaper questionnaire, a door-to-door survey, and 3 laboratory experiments, the authors examined a proposed effect of shared participation in novel and arousing activities on experienced relationship quality. The questionnaire and survey studies found predicted correlations of reported shared "exciting" activities and relationship satisfaction plus their predicted mediation by relationship boredom. In all 3 experiments, the authors found predicted greater increases in experienced relationship quality from before to after participating together in a 7-min novel and arousing (vs. a more mundane) task. Comparison with a no-activity control showed the effect was due to the novel-arousing task. The same effect was found on ratings of videotaped discussions before and after the experimental task. Finally, all results remained after controlling for relationship social desirability. Results bear on general issues of boredom and excitement in relationships and the role of such processes in understanding the typical early decline of relationship quality after the honeymoon period.
Journal of Personality and Social Psychology
2000,
Vol. 78, No. 2, 273-284
Copyright 2000 by the American Psychological Association, Inc.
0022-3514/00/J5.00 DOI: 10.1037//0022-3514.78.2.273
Couples' Shared Participation in Novel and Arousing Activities and
Experienced Relationship Quality
Arthur Aron
State University of New York at Stony Brook
Christina C. Norman
Fountain House Inc.
Elaine N. Aron
State University of New York at Stony Brook
Colin McKenna
Kaiser Permanente Medical Center
Richard E. Heyman
State University of New York at Stony Brook
Using a newspaper questionnaire, a door-to-door survey, and 3 laboratory experiments, the authors
examined a proposed effect of shared participation in novel and arousing activities on experienced
relationship quality. The questionnaire and survey studies found predicted correlations of reported shared
"exciting" activities and relationship satisfaction plus their predicted mediation by relationship boredom.
In all 3 experiments, the authors found predicted greater increases in experienced relationship quality
from before to after participating together in a 7-min novel and arousing (vs. a more mundane) task.
Comparison with a no-activity control showed the effect was due to the novel-arousing task. The same
effect was found on ratings of videotaped discussions before and after the experimental task. Finally, all
results remained after controlling for relationship social desirability. Results bear on general issues of
boredom and excitement in relationships and the role of such processes in understanding the typical early
decline of relationship quality after the honeymoon period.
This article focuses on a proposed effect on relationship quality
of shared participation in novel and arousing activities (over and
above any effect of shared participation in more mundane activi-
ties).
This proposed effect bears on basic motivational and cogni-
tive mechanisms underlying excitement and boredom in long-term
relationships and has significant potential practical applications.
We first briefly explore the background and preliminary re-
search suggesting that such an effect should occur. We then
present data from two questionnaire studies, each of which show
the predicted pattern of associations, and then a series of three
experiments, which demonstrate the effect under controlled labo-
ratory conditions and establish an innovative practical paradigm
for future research exploring more deeply the underlying
mechanisms.
Arthur Aron, Elaine N. Aron, and Richard E. Heyman, Department of
Psychology, State University of New York at Stony Brook; Christina C.
Norman, Fountain House Inc., New York; Colin McKenna, Kaiser Perma-
nente Medical Center, Vallejo, CA.
This research was funded in part by National Science Foundation Grant
SBR9514417. We thank the following individuals for their help with this
research: Laura Adman, Abigail Arevalo, Lorraine Bifulco, Megan Dlugo-
zima, Shari Feldbau-Kohn, Valerie Galati, Helen Hill, Lata Jani, Yoon
Kim, Debra Mashek, Laura Palmer, Candy Raphaiel, Pamela Schlossman,
Jayne Schneider, Erica Shertzer, and Mark Skellie.
Correspondence concerning this article should be addressed to
Arthur Aron, Department of Psychology, State University of New York,
Stony Brook, New York 11794-2500. Electronic mail may be sent to
aron@psychl .psy.sunysb.edu.
North American couples clearly consider spending time to-
gether, regardless of the type of activity, to be an important
maintenance strategy (Baxter & Dindia, 1990; Dindia & Baxter,
1987).
There is also substantial evidence that time spent together
is indeed correlated with relationship quality. For example,
signif-
icant associations were found in five separate U.S. studies con-
ducted in the last 30 years using probability samples (Kilbourne,
Howell, & England, 1990; Kingston & Nock, 1987; Orden &
Bradburn, 1968; Orthner, 1975; White, 1983). Several studies
(Holman & Jacquart, 1988; Kingston & Nock, 1987; Orden &
Bradburn, 1968; Orthner, 1975) reported substantially stronger
correlations with relationship quality for activities that were in-
tensely interactive versus passive, parallel, or merely in the com-
pany of
others.
Hill (1988), in finding a strong overall link between
shared activities and marital stability, reported the strongest effects
for shared "recreational activities," all of which were somewhat
active and arousing (such as "outdoor activities, active sports, card
games, and travel," p. 447). However, to our knowledge, other
than the preliminary study described below (Reissman, Aron, &
Bergen, 1993), there have been no previous studies directly focus-
ing on the proposed effect on relationship quality of shared par-
ticipation in novel and arousing activities on marital quality.
The proposed effect on relationship quality of shared participa-
tion in novel and arousing activities first suggested itself to us in
the context of our self-expansion model and its connection to
general issues of excitement and boredom in close relationships
(Aron & Aron, 1986). The occasional explanations that have been
offered for the well-documented precipitous decline in relationship
satisfaction after the early relationship years (e.g., Blood & Wolfe,
273
274
ARON,
NORMAN,
ARON,
McKENNA,
AND HEYMAN
1960;
Glenn, 1990; Locke & Wallace, 1959; Rollins & Feldman,
1970;
Tucker & Aron, 1993) have mostly invoked notions of habit-
uation and optimal level of arousal (Aron & Aron, 1986,1997; Aron,
Aron, & Norman, in press; Aronson & Linder, 1965; Berger, 1988;
Huesmann, 1980; Jacobson & Margolin, 1979; Livingston, 1980;
Plutchik, 1967).' In this light, we reasoned that the initial exhilaration
in the early relationship years may be due to the novelty and arousal
of forming the relationship; but when, over time, this novelty and
arousal inevitably decline, couples might instead maintain a high level
of experienced relationship quality by engaging together in novel and
arousing activities, so that the positive effect from such activities
becomes associated with the relationship. Indeed, couples often do
seem to adopt this strategy of engaging jointly in expanding activi-
ties—in many cultures, traditionally they build a home and family
(although in North American culture these are not always shared or
central, or may be a source of stress and too much novelty and arousal,
as suggested by the apparently negative impact of the birth of the first
child; see, e.g., Belsky, 1985; Tucker & Aron, 1993). Other examples
in North American culture are causes taken on jointly, businesses run
together, and shared professional or recreational activities. It is pos-
sible that these kinds of shared experiences provide relationship
satisfaction because they are intrinsically enjoyable and by being
shared become associated with the partner and the relationship.
There are also several other mechanisms through which shared
participation in novel and arousing activities might enhance experi-
enced relationship quality. First, shared novel and arousing activities
might be expected to generate more positive feelings toward the
partner through the same processes associated with the arousal-
attraction effect (e.g., Dutton & Aron, 1974), such as misattribution of
arousal, negative reinforcement, or facilitation of the dominant re-
sponse (for a recent review, see Foster, Witcher, Campbell, & Green,
1998).
Another mechanism that might create the proposed effect is
that most such activities in which a couple might engage together
would also be likely to involve cooperation, thus reinforcing a sense
of interdependence and closeness. Finally, it is possible that aside
from the shared aspect, just participating in such activities produces an
overall increase in positive evaluation of all aspects of one's experi-
ence,
so that one's relationship might also be seen more positively.
Thus,
there are a number of interesting lines of understanding that
might be marshaled as a foundation for our prediction. (Indeed, we
hope that demonstrating the basic effect and establishing a practical
laboratory paradigm for studying it—the purposes of the present
article—will lead to further research specifically designed to distin-
guish among possible mechanisms.)
At the same time, the various theoretical arguments in support of
the proposed effect are not so strong as to make empirical demon-
stration unnecessary. Indeed, there are also theoretical reasons not to
expect the proposed effect. First, the existing lines of research seem to
imply that the major sources of relationship quality are either pre-
existing (e.g., neuroticism, attachment working models, gender, com-
munication skills, similarity of values), not systematically associated
with declines over time (e.g., establishment of trust, transformation of
motivation, investments, decline in alternatives), or extrinsic to the
relationship (e.g., income, birth of child, and other external stressors).
Taken together, these variables do not seem to leave much to be
accounted for by the proposed phenomenon. Second, it is plausible
that shared participation in novel and arousing activities could under-
mine relationship quality—such participation might just add to exist-
ing stressors. Or, spending time together in novel and arousing activ-
ities could force partners to attend to each other under circumstances
where the partner might behave less than optimally (given the novelty
and arousal associated with it). Another possibility is that such activ-
ities,
to the extent they create feelings similar to the initial relationship
stages of falling in love (but not as strong), may remind the partners
that they no longer feel the same intensity about the partner or
relationship. Still another way such activities might make things
worse is that if there are ongoing sources of conflict, the increased
arousal could easily be mislabeled as distress or anger. In fact, trying
to carry out a novel and arousing activity together could well create
conflict because these are circumstances in which couples are unlikely
to be well coordinated and in which they will not have learned how to
accommodate to their differences.
Thus,
although there are various theoretical reasons for hypoth-
esizing the proposed effect, there are sufficient reasons not to
expect it (or to predict its opposite) so that empirical research
remains quite necessary.
Preliminary Evidence
On the basis of the various arguments presented above for
expecting such an effect, it seemed appropriate to conduct a
preliminary study. Reissman et al. (1993) recruited 53 upper-
middle class, middle-aged married couples in Palo Alto, Califor-
nia, who agreed to take part in a 10-week study. Participants were
randomly assigned to one of three conditions. In one condition, the
couple was instructed to spend 1.5 hr per week doing an activity
from a list of "exciting activities" provided to them. This list
consisted of activities both partners had independently rated as
highly exciting on a prestudy questionnaire that included a large
list of couple activities. Couples in a second condition were given
a list of activities both had rated as highly "pleasant" but not as
"exciting." (Examples of exciting activities on these lists which
were actually carried out included attending musical concerts,
plays,
and lectures; skiing; hiking; and going dancing. Some
examples of the pleasant activities were visiting friends, attending
a movie, attending church, and eating out.) Couples in the third
condition were a no-activity control group. All couples completed
a standard marital quality questionnaire and a relationship-relevant
social desirability scale at the start of the study and after 10 weeks.
The key finding was a significantly greater increase in satisfaction
in the exciting activities group than in the pleasant activities group,
an effect which remained even after partialling out changes in a
measure of relationship-relevant social desirability response bias.
Thus,
this result supported the hypothesized effect that shared
participation in novel and arousing activities (in the Reissman et al.
study operationalized as exciting) increases marital quality.
However, these results were not entirely unambiguous. The
contrast of the exciting condition with the no-activity control (for
which change in satisfaction was intermediate between the two
1
A related line of thinking comes from psychodynamic theories of
idealization, which discuss the decline in terms of increasing familiarity,
making it more difficult to project an all-loving parent (Bergler, 1946),
ego-ideal (Reik, 1944), or anima/animus (Jung, 1925/1959) onto the other
in the relationship. Recent research by Murray, Holmes, and Griffin (1996)
supports the importance of idealization as a source of high relationship
quality; however, their data suggest that the effect is long-term and does
not create a precipitous decline after the initial relationship period.
NOVEL-AROUSING ACTIVITIES AND RELATIONSHIP QUALITY
275
activity conditions) was in the expected direction, but not
signif-
icant. Of course, this was a small sample study with a fairly slight
intervention into people's lives of 1.5 hr per week. In addition,
being a field study, it was subject to large amounts of uncontrolled
variance in the form of other influences on people's lives. Thus,
although this study provided encouraging initial support for the
proposed effect, it is hardly conclusive. Further, although the
research paradigm used in the Reissman et al. (1993) study has
great advantages in terms of external validity, it is not highly
controlled and is not a very practical paradigm for future research
focusing on mechanisms. (By contrast, the paradigm used in the
experiments reported in this article is much more highly con-
trolled—it is a laboratory procedure—and is considerably more
practical as a model for future research on this topic.) Finally, the
Reissman et al. experiment provides only limited support for the
proposed effect because it tested only the most general hypothesis
and because its sample was particularly unrepresentative (for ex-
ample, 74% of wives were college graduates, most with at least
some postgraduate education). In sum, the Reissman et al. study
was encouraging, but further research was still needed to demon-
strate the effect.
The Present Research
We conducted two questionnaire studies and three laboratory
experiments to test the basic effect and to explore specific hypoth-
eses associated with the effect; we also intended for the laboratory
experiments to illustrate a practical and rigorous research para-
digm for future studies on this effect.
The first two studies included data from a newspaper questionnaire
and from a door-to-door survey. These two studies permitted corre-
lational tests of the hypothesized link between shared participation in
novel and arousing activities and experienced relationship quality. In
addition to the basic correlation, they examined an additional predic-
tion from the general line of reasoning we presented
earlier:
The basic
association would be mediated by perceiving the relationship overall
not to be boring. (The logic is that engaging in exciting activities
makes one perceive the relationship as not boring, and this leads to
enhanced relationship quality.) Thus, we tested the following two
hypotheses in each questionnaire study:
Hypothesis
1: Shared participation in novel-arousing activities is
associated with higher levels of experienced relationship quality.
Hypothesis
2:
This association is mediated by the extent to which the
relationship is perceived as boring (versus exciting).
Studies 3-5 were experiments that all used the same basic
paradigm. Couples were recruited to participate in what they
believed was an evaluation session in which they would complete
some questionnaires, would be videotaped participating in a phys-
ical activity, and would then complete some additional question-
naires. In fact, the first set of questionnaires included baseline
measures of experienced relationship quality; the physical activity
was experimentally manipulated to be either novel and arousing or
mundane (i.e., pleasant but not arousing and less novel); and the
final set of questionnaires included the posttest measures of expe-
rienced relationship quality. In Study 3, we tested the basic effect
with a sample of mainly dating or living-together undergraduate
couples. This initial experiment tested bur basic causal hypothesis
that shared participation in novel and arousing activities enhances
experienced relationship quality. In Study 4, we replicated Study 3,
but also included a no-activity control condition and used longer
term, mainly married couples recruited from the community. In
Study 5, we used a sample of community couples and included
only the two key conditions of Study 3. Also, Study 5 couples were
videotaped in verbal interactions before and after the experimen-
tally manipulated interaction activity. The videotapes, which were
then coded according to a standard protocol, provided the oppor-
tunity to assess whether the hypothesized effect would also be
observed with these more objective indicators of experienced
relationship quality. This addition is particularly important because
it examines whether the proposed effect goes beyond subjective
experience, thus minimizing the possibility that other results of
these studies are somehow due to some limitation of self-report
measures. In sum, we tested the following hypotheses in Studies 3-5:
Hypothesis 3. Shared participation in novel-arousing activities, com-
pared with shared participation in mundane activities, increases ex-
perienced relationship quality. (Tested in Studies 3—5.)
Hypothesis 4: Shared participation in novel-arousing activities, com-
pared with a no-activity control condition, increases experienced
relationship quality. (Tested in Study 4.)
Hypothesis 5: Shared participation in novel-arousing activities, com-
pared with shared participation in mundane activities, increases rela-
tionship quality as measured by systematically coded standardized
verbal interactions. (Tested in Study 5.)
In considering these hypotheses we were quite aware that mea-
sures of relationship experience have often been found to be
influenced by a tendency for socially desirable responding (see,
e.g., Anderson, Russell, & Schumm, 1983; Edmonds, 1967; Ed-
monds, Withers, & Dibatista, 1972). That is, as is often the case
with self-report measures generally, responses to questionnaires
about relationship quality are likely to be biased by individual
differences in tendencies of self-deception and impression man-
agement (Paulhus, 1984), presumably serving needs to see oneself
and to be seen by others in the best possible light. Thus, in all five
studies we included measures of the tendency for socially desirable
responding in a relationship context and conducted an additional
test of each hypothesis controlling for this measure.
Study 1
Method
As part of a larger study (McKenna, 1989), we arranged to print a
marital questionnaire in the Santa Cruz, California, Sentinel, a small local
newspaper.
2
Santa Cruz is a university and resort community with a mostly
2
The newspaper survey was originally conducted to examine associa-
tions of personality variables with relationship quality; results of analyses
conducted for that purpose were reported in a dissertation (McKenna,
1989).
Study 2 was conducted as a replication of that dissertation but did
not yield consistent results in the personality analysis. Neither study has
been published or submitted elsewhere for publication. The present use of
these data was made possible by these two studies having included,
serendipitously, in addition to measures of experienced relationship quality
and relationship social desirability, two items on relationship boredom and
a key item on the amount of "exciting" activities participated in with the
partner.
276
ARON,
NORMAN,
ARON,
McKENNA,
AND HEYMAN
White, lower-middle
and
middle-class population.
Of the 178
complete
responses received within
a
week
of
its publication, 112 met our criteria of
being married
or
living together with their partner and
in a
relationship
of
no more than
15
years.
3
These included
90
women
and 22 men;
mean
relationship length was 6.37 years,
and
mean age was 38.82. Respondents
were instructed
to cut
the questionnaire out
of
the paper and mail
it in (for
a discussion
of the
strengths
and
limitations
of
newspaper surveys,
see
Shaver
&
Rubenstein, 1983).
The measure
of
experienced relationship quality was the 10-item Dyadic
Satisfaction subscale
of
Spanier's (1976) widely used Dyadic Adjustment
Scale.
An
example question
on the
subscale
is, "In
general,
how
often
do
you think that things between
you and
your partner
are
going well?"
answered
on a
6-point scale ranging from
1 (all of
the time) through
6
(never).
We
used
the
single subscale rather than
the
full measure because
of the need
to
keep
the
number
of
items
to a
minimum. We also think that
satisfaction
is
more central
to our
notion
of
what should
be
influenced
by
shared novel
and
arousing activities (what
we are
calling experienced
relationship quality) than
are the
other subscales—Dyadic Differences,
Interpersonal Tensions and Personal Anxiety, and Dyadic Cohesion. Alpha
in our sample was .87. Our measure
of
relationship social desirability
was
Olson, Fournier, and Druckman's (1982) 5-item Idealistic Distortion Scale.
This
is a
shortened version
of
Edmonds's (1967) Marital Conventionaliza-
tion Scale,
in
which
the
selected items were also slightly modified
to be
applicable
to
nonmarried partners.
All
items
are
answered according
to a
true-false format.
An
example item is, "My partner and
I
understand each
other completely." Alpha
in our
sample
was
.77.
We
also constructed
a
two-item measure
of
boredom with
the
relationship: "How bored
are you
with your current relationship?"
and
"How exciting
is
your current rela-
tionship?" (reverse scored). Each item
was
answered
on a
5-point scale
(e.g.,
1 = Not
bored
at
all,
5 =
Extremely bored). Alpha was .79. Finally,
the questionnaire asked, "How exciting are the things you do together with
your partner?" answered
on a
5-point scale ranging from 1 (Not exciting
at
all)
to 5
(Extremely exciting).
As
noted earlier, "exciting" was intended
as
an operationalization
in
ordinary language
of
novel
and
arousing.
Of
course,
it
might have been preferable
to
measure this
key
variable using
more than
a
single item. However,
any
limitation
due to low
reliability
works against
our
hypotheses
(it
would directly undermine Hypothesis
1,
and Hypothesis
2
cannot
be
tested
if
Hypothesis
1 is
nonsignificant).
Results
Regarding Hypothesis 1, as predicted, there was a strong posi-
tive association between responses to the exciting activities ques-
tion and experienced relationship quality (r =
.51,
p < .001).
4
'
5
This association remained significant, though was somewhat
smaller, after controlling for relationship social desirability (r =
.32,
p < .001). Thus, Hypothesis 1 seems to be clearly supported,
even when controlling for relationship-relevant social desirability.
Hypothesis 2 was that this association would be mediated by
boredom with the relationship. The results clearly support this
hypothesis. As shown in Figure 1, the beta of
.51
between exciting
activities and experienced relationship quality became
nonsignif-
icant and dropped to a beta of .10 (ns) when boredom with the
relationship was included in the model. At the same time, boredom
with the relationship, which correlated —.56 with exciting activi-
ties and
.79 with quality, retained a significant beta of
.74 (p <
.001) with quality when the exciting activities variable was in-
cluded in the model. As a direct test of the mediation, using the
version of Sober
s
(1982) formula recommended by Kenny,
Kashy, and Bolger (1998), we tested the significance of the com-
pound path from exciting activities to relationship boredom to
quality in the context of the model that included the direct path
from activities to quality. This test was clearly significant
(Z = 5.81, p < .001).
We also repeated the entire mediation analysis partialing out
relationship social desirability. The mediation was again sup-
ported: The correlation between quality and exciting activities of
.32 dropped to a nonsignificant beta of .04, but the boredom-
quality association, which was originally —.67, remained a sub-
stantial —.60 (p < .001) when the exciting activities variable was
included in the equation (Sobel's test, Z = 4.17, p < .001).
3
We limited
our
sample
to
participants with
a
maximum
of 15
years
together
in
order
to be
consistent with
the
criteria
we set
when recruiting
community couples
for
Studies
4
and 5, where longer term couples tended
to
be
older and thus often problematic
for
the physical tasks.
In
this study,
we also
ran
analyses parallel
to
those reported here including
all 178
participants (some
of
whom reported being married
for as
long
as 60
years!).
Results were nearly identical
in
each case, except that with
the
larger sample size, results reached even more stringent levels
of
signifi-
cance. Also,
in
both questionnaire studies there were
a few
cases
in
which
we were aware that
we had
responses from both members
of a
couple
(in
Study
1, the
responses arrived
in the
same envelope;
in
Study
2,
both
members
of
the couple completed
the
questionnaire during
the
interview-
er's visit).
To
examine whether
our
results might
be
inflated
by
noninde-
pendence, we reconducted
all the
analyses
for
Studies 1 and
2
treating
the
scores from known couples
as
single scores (couple means). Results
of
these analyses were virtually identical
to
those treating
all
individuals
as
independent.
4
Throughout this article,
all
analyses
are
conducted partialing
out
rela-
tionship length
(in
addition
to
any other variables that might
be
controlled
in
a
particular analysis
and
noted
in the
text
as
such). This
was
done
because
of the
substantial literature, noted previously, linking length
to
(decreasing) relationship satisfaction.
In
most cases, controlling
for
length
made little difference. For example, the present correlation not controlling
for relationship length
is
nearly identical (.50). Some (e.g., Aron, Aron,
&
Smollan, 1992) have argued that the psychological meaning
of
relationship
length
may
actually
be
more closely related
to the log of
actual length.
Thus,
we also conducted parallel analyses using log
of
relationship length.
Results were nearly identical
to
those using actual relationship length.
For
simplicity, we report only results controlling
for
direct relationship length.
Also,
in
each analysis conducted throughout this article,
we
tested
for
interaction with gender.
In
this study,
the
power
of
gender
was
only
modest. However,
in
Study
2
the gender distribution was nearly equal
and
in
the
experiments (Studies
3-5),
where gender was
a
repeated measures
variable,
the
power
for
testing
for
gender interactions
was
quite high.
Nevertheless, there were fewer gender interactions than would be expected
by chance, and the few that showed up did not show any consistent pattern
across
or
within studies. Thus,
all
results
of the
surveys combine
all
individuals
in the
sample
and all
data from
the
experiments use couple
as
the unit
of
analysis (with couple averages
on
each variable
as the
data
points).
5
This tendency was entirely monotonic throughout the measured levels
of exciting activities. However,
we did
find
a
small,
but
significant,
tendency
for a
quadratic effect
in
which the overall positive association
of
exciting activities with relationship quality becomes weaker
at
higher
levels
of
exciting activities. This
is
consistent with
an
optimal level
of
arousal (Berlyne, 1960) interpretation
of
the effect. This pattern
of a
slight
curvilinearity was found
in
this study both not controlling
for
and control-
ling
for
relationship social desirability. However, there
was no
parallel
tendency whatsoever
in
any
of
the analyses
of
the Study
2
data.
In
fact,
the
raw coefficients were slightly
in
the direction
of
an increase
in
the positive
effect
at
higher levels
of
exciting activities.
NOVEL-AROUSING ACTIVITIES AND RELATIONSHIP QUALITY
277
Novel/Arousing
Activities
Relationship
Quality
Novel/Arousing
Activities
_.LQ
-.56***
-.74***
Figure 1. Results for the test of Hypothesis 2 based on newspaper survey data (Study 1; N = 112). The path
diagrams show mediation by reported relationship boredom of the association between novel-arousing activities
(reported participation in "exciting" activities) and experienced relationship quality (scpres on the Dyadic
Satisfaction subscale of Spanier's, 1976, Dyadic Adjustment Scale). Path coefficients shown are standardized
regression coefficients. *** p < .001.
One aspect of the mediation results also argues against the
correlations among our variables as simply due to a general ten-
dency to experience everything about the relationship as similarly
positive or negative. If this were the case, partialing out reported
exciting activities should have eliminated the association of bore-
dom with experienced relationship quality (which it did not). This
same result also argues against the specific concern that partici-
pating in exciting activities and experiencing relationship boredom
are merely two different operationalizations of the same underly-
ing construct. If that were the case, then including both as predic-
tors in the same equation would render both null (which it did not).
Finally, the fact that the boredom-quality association remained
after partialing out exciting activities is a further argument (in
addition to the fact that results hold up after controlling for social
desirability) against our associations being merely an artifact of
common method variance.
Study 2
Method
We conducted a door-to-door survey in which a female interviewer
approached houses and apartments selected on a random basis from two
neighborhoods in Santa Cruz. Eligible participants (being married or living
with their partner for 15 years or less) were asked to complete the 10-min
questionnaire on a clipboard while the interviewer waited. Approximately
70%
of eligible participants who were at home completed the question-
naire. The final sample of 80 individuals included 42 women and 38 men;
had a mean age of 30.96 and a mean relationship length of 4.60 (ranging
from 2 months to 11 years). The questionnaire included the same items as
were used in the newspaper survey (Study 1). Alphas in this sample were
.78 for experienced relationship quality, .75 for boredom with the relation-
ship,
and .68 for relationship social desirability.
Results
Regarding Hypothesis 1, that shared participation in novel and
arousing activities is associated with experienced relationship
quality, there was a strong positive beta of .45 (p < .001), which
remained significant though somewhat smaller (/3 =
.24,
p < .01)
after controlling for relationship social desirability.
Results for Hypothesis 2, that the association would be at least
partially mediated by boredom with the relationship, are shown in
Figure 2. As can be seen from the figure, we again found the
predicted pattern. The beta of .45 between exciting activities and
experienced relationship quality dropped to a beta of .22 when
boredom with the relationship was added to the equation. At the
same time, boredom with the relationship, which correlated
.52
with exciting activities and —.58 with relationship quality retained
a strong beta of —.44 (p < .001) with relationship quality when
exciting activities was included in the equation. Overall, the me-
diation was clearly significant; Sobel's test, Z = 3.38,p <
.001.
A
parallel mediation analysis partialing out relationship social desir-
ability also clearly supported the predicted mediation pattern: The
exciting-activities-relationship-quality beta of .24 dropped to a
nonsignificant beta of .12; but the boredom-quality beta, which
was originally -.36, remained at a beta of -.31 (p < .01) when
exciting activities were included in the equation; Sobel's test,
Z= 2.52, p< .01.
Discussion of Studies 1 and 2
Studies 1 and 2, using diverse samples and methods, were
entirely consistent in supporting a positive association between
participation in novel and arousing activities and experienced
relationship quality and a mediation of this association by per-
ceived boredom in the relationship. Further, all of these findings
held up quite nicely even after controlling for a standardized,
reasonably reliable measure of relationship-relevant social desir-
ability response bias. It is important to emphasize that both this
result and the mediation findings provide substantial evidence that
the basic pattern of association we observed between exciting
activities and relationship quality was not merely an artifact of
278
ARON,
NORMAN,
ARON,
McKENNA,
AND HEYMAN
Novel/Arousing
Activities
.45***
Relationship
Quality
Novel/Arousing
Activities
,22.
-.51*
-.44***
Figure 2. Results for the test of Hypothesis 2 based on door-to-door survey data (Study 2; N = 80). The path
diagrams show partial mediation by reported relationship boredom of the association between novel-arousing
activities (reported participation in "exciting" activities) and experienced relationship quality (scores on the
Dyadic Satisfaction subscale of Spanier's, 1976, Dyadic Adjustment Scale). Path coefficients shown are
standardized regression coefficients.
*
p < .05.
***
p < .001.
measuring all the variables with similar questionnaire items or of
a general tendency to say good things about the marriage.
Nevertheless, Studies 1 and 2 were correlational and cross-
sectional, so that any conclusions regarding direction of causality
are necessarily limited. Thus, we next conducted a series of ex-
periments (Studies 3-5) to help sort out directions of causality
regarding the basic effect, as well as to test additional hypotheses.
The purpose of the experiments was also to develop a practical
laboratory paradigm to permit future research to sort out mecha-
nisms and other issues regarding the basic effect.
Study 3
In this experiment we first developed our laboratory paradigm
and applied it to test the basic effect, permitting us to test our key
prediction about direction of causality (Hypothesis 3).
Method
Participants. Participants were 28 couples (24 dating and 4 married)
who had been in their relationship from 2 months to 8 years (M = 2.64
years);
ages ranged from 17 to 44 (M = 23.13). (Four additional couples
were not included in the analyses because they did not complete the task
properly; they included 1 couple in the novel-arousing-activity condition
and 3 couples in the mundane-activity condition.) Couples were recruited
from psychology classes and by advertisements on flyers posted through-
out the campus of the State University of New York at Stony Brook.
Participants recruited from psychology classes partially fulfilled a class
research participation requirement; participants recruited through the
posters were paid $10 per couple. Participants were recruited for an
experiment on "factors that affect relationships." When prospective par-
ticipants were called for appointments, they were screened for any medical
condition that would prevent them from performing physical or aerobic
activity. They were also advised at this time that they should dress
comfortably.
Procedure. Each couple was tested individually. When the couple
arrived at the laboratory, they were greeted by a female experimenter who
was unaware of the experimental hypothesis. First, the partners were taken
to separate rooms to fill out the pretest questionnaire. Second, they per-
formed a task together (described below), which they believed was an
evaluative task. Third, they filled out the posttest questionnaires, in sepa-
rate rooms; and fourth, they were debriefed by another experimenter (who
was aware of the experimental hypothesis). To maximize statistical power,
we measured experienced relationship quality both before and after the
experimental manipulation (the task), with the pretest serving as a covariate
to reduce the influence of random variance over couples.
Pretest index of experienced relationship quality. In order to capture a
wider range of content of experienced relationship quality than we had in
the surveys, at both pretest and posttest we constructed measures that
included indicators of both relationship satisfaction and passionate love.
We felt it was especially important to go beyond the usual sole reliance on
satisfaction in the present context of issues of boredom, arousal, and
excitement. For the pretest, we assessed the satisfaction aspect using
Hendrick's (1988) Relationship Assessment Scale, and we assessed the
passionate love aspect using Hatfield and Sprecher's (1986) Passionate
Love Scale. The Relationship Assessment Scale is a 7-item generic mea-
sure of relationship satisfaction that participants completed using a 7-point
scale whose anchors varied slightly from item to item. Example items are
"How well does your partner meet your needs? (1 = Not very well, 1 =
Very well) and "How good is your relationship compared to most?" (1 =
Not good at
all,
7 = Very good). Hendrick reported an alpha of
.87;
in the
present study, alpha was .88. The Passionate Love Scale (short form) is a
15-item measure of thoughts, feelings, and behaviors characteristic of
passionate love that our participants completed using a 6-point scale
ranging from 1 (Untrue) to 6 (True). Example items are "Since I've been
involved with my partner, my emotions have been on a rollercoaster" and
"I feel happy when I am doing something to make my partner happy."
Hatfield and Sprecher examined the scale's reliability and validity in a
study of undergraduates in dating or more serious relationships. The alpha
was .91 for the short version of the scale. Validity was supported for the
scale by significant correlations of it with other love measures, as well as
with measures of commitment, satisfaction with the overall relationship,
and satisfaction with the sexual aspects of the relationship. We chose to use
this scale (and the posttest measure of passionate love described below) in
our study because it is one of the few measures of love that is not based on
global questions or focused on companionate-type love. In our sample,
alpha was .83, and the correlation between these satisfaction and love
NOVEL-AROUSING ACTIVITIES AND RELATIONSHIP QUALITY
279
components of our pretest measure was .51. We created our combined
pretest index of experienced relationship quality by averaging Z scores on
the two scales.
6
Posttest index of experienced relationship quality. To avoid partici-
pants'
recalling responses from the pretest (and also to avoid participants
realizing that the second set of questionnaires was a posttest), it was necessary
to use different measures for the posttest. For the satisfaction aspect of our
posttest index, we used Huston, McHale, and Crouter's (1986) Marital Opin-
ion Questionnaire; for the passionate love aspect, we used Mathes's (1982)
Romantic Love Symptom Check List. The Marital Opinion Questionnaire is a
10-item semantic differential scale plus one global question on overall rela-
tionship satisfaction. Respondents characterize their relationship in terms of a
series of bipolar adjectives such as "miserable-enjoyable." Four points sepa-
rate the two endpoints (making a 6-point scale). The global question asks "All
things considered, how satisfied or dissatisfied have you been with your
relationship over the past TWO months, with 1 being completely dissatisfied
and 6 being completely satisfied?" Belsky, Spanier, and Rovine (1983) re-
ported the reliability of this scale to range from .87 to .94 with a mean of .91
across three times of measurement. In our sample, alpha was .95. The Ro-
mantic Love Symptom Check List consists of 35 feelings associated with
romantic love, to each of which respondents indicate whether it applies to them
when they think of their beloved. Items include "tingling," "bursting with
happiness," and "an increased heartbeat." Mathes reported an alpha of
.95
and
strong correlations with other love measures. The correlation between these
satisfaction and love measures that comprised our posttest index of experi-
enced relationship quality was .43 and we created the index averaging Z scores
on the two scales.
Relationship social desirability. For this experiment, we split the 15
items of the short form of Edmonds's (1967) Marital Conventionalization
Scale into two parts, with 7 items administered at pretest (a = .79) and 8
items administered at posttest (a = .72).
Manipulation check. We evaluated the effectiveness of the manipula-
tion with a 5-item index asking how fun, interesting, exciting, boring
(reverse scored), or dull (reverse scored) participants thought the task had
been, each rated on a 10-point scale ranging from 1 {not at all) to 10 (very
much).
Alpha was .93. The manipulation check items were administered
after completion of all other measures and just before the debriefing.
Manipulation of task type (novel—arousing vs. mundane). Couples
were randomly assigned to either a novel-arousing or mundane task
condition. We designed the novel—arousing task to involve many novel
elements and to be physiologically arousing. Gymnasium mats were set up
across the diagonal (approximately 9 m) of a large room in our laboratory.
In the center of the room, across the mats, we laid a rolled-up gym mat
(approximately 1 m high) as a barrier for the participants to cross over.
Before beginning, participants in the novel-arousing condition were bound
with their partner on one side at the wrist and ankle with Velcro straps.
(This also provided a rationale for asking them to remove any watches.)
They were then instructed to travel the length of the mats (remaining on
their hands and knees at all times) and return to the starting point, crossing
the barrier once in each direction. In addition, couples were required to
carry a cylindrical pillow down and back with them, without using hands,
arms,
or teeth—even while going over the barrier. (This could only be
done by holding it between their heads or bodies.) Couples were told that
they would be given four trials to complete the task in under 1 min (in
which case they would win a small prize of candy), and that only approx-
imately 25% of the couples who had participated in the experiment thus far
had been successful. After the third trial, most couples were told that they
had successfully completed the course in under 1 min (regardless of
whether they actually had). However, if a couple had performed unusually
poorly on one of the first thee trials, so that it would not be believable to
say they had completed the course in under
1
min, a fourth trial was given,
and this was described as a successful trial. (In all cases where a fourth trial
was needed, the experimenter felt that it would be believable that it had
been completed in under 1 min.)
The mundane task was designed to be as similar as possible but less novel
and arousing. This activity also took place on the gym mats arranged in the
same way. However, the activity consisted of Partner
1
(randomly determined)
rolling a ball to the center of the room (the barrier) while on his or her hands
and knees, while Partner 2 remained at the edge of the room. Once Partner 1
had reached the center, Partner 2 crawled toward the center, retrieved the ball
from Partner 1, and then returned to the edge of
the
room where she or he was
originally situated. Once Partner 2 reached the edge, Partner 1 also was
permitted to return to the edge (opposite Partner 2). This cycle was repeated
for 7 min. Partners were told they were expected to coordinate their activity
and to work slowly. To further ensure that participants carried out the task
slowly, we sounded a beep every second and the participant was instructed to
make only one move of an arm or leg at each beep.
Results and Discussion
Scores on the manipulation check items were significantly
higher (more exciting, fun, interesting, etc.) in the novel-arousing
condition (M = 8.77) than in the mundane condition (M = 6.08),
F(l,
25) = 25.45, p < .001, r (effect size) = .71.
Hypothesis 3 was that shared participation in novel-arousing
activities, compared with shared participation in mundane activi-
ties,
increases experienced relationship quality. To test this hy-
pothesis, we conducted an analysis of covariance (ANCOVA)
comparing the two experimental groups on couple average posttest
experienced relationship quality, with couple average pretest ex-
perienced relationship quality (and relationship length) as a co-
variate.
7
Hypothesis 3 was clearly supported, F(l, 24) =
6.07,
p <
.05,
partial r = .45. The adjusted means on the posttest experi-
6
We also included a variety of other personality and relationship mea-
sures (e.g., closeness, commitment, perceived alternatives) at pretest and
posttest for use in another study relating personality to relationship char-
acteristics. That research program is still in progress and no papers have
been submitted from it as yet.
7
The pooled within-group correlation (i.e., the correlations partialing out
condition) of pretest and posttest composites was
.74.
This lends support
to
our
approach of using these different scales as parallel pretest and posttest indi-
cators. (Pooled within-group pretest-posttest correlations in Studies 4 and 5
were .81 and .79, respectively.) The use of a pretest-posttest design co