journal of Personality and Social Psychology Copy-right 198? by the American Psychological Association, Inc.

1987. Vol. 53. No. 2, 397-410 0022-3SI4/87/S00.75

Maintenance of Satisfaction in Romantic Relationships:

Empathy and Relational Competence

Mark H. Davis and H. Alan Oathout

Eastern Illinois University

On the basis of aseries of recent investigations linking personality, social behavior, and social satisfac-
tion, as well as the theoretical concept of relational competence, we propose a model of relationship
satisfaction. The mode! is based on the notion that personality in general and empathy in particular
affect relationship satifaction through their influences on specific mediating behaviors. The tenets
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

of the model were tested by assessing the key constructs for both members of 264 heterosexual
This document is copyrighted by the American Psychological Association or one of its allied publishers.

romantic couples. In general, the model was strongly supported, as three separate facets of disposi-
tional empathy had separate and predictable influences on self-reported behavior, which in turn
significantly influenced partners' perceptions of those behaviors. Also as expected, perceptions of
partner behavior were significant influences on one's satisfaction with the relationship. The model
worked especially well for longer term relationships and somewhat better for predicting female be-
havior; the role of one facet of empathy—perspective taking—was especially strong for longer rela-
tionships. Thus, the model appears to be a fruitful way to examine the role of personality on social
and psychological well-being,

In what fashion does personality affect social and psychologi- in turn negatively associated with loneliness (Davis, Franzoi, &
cal well-being? Although frequently posed, this question has Wellinger, 1985). Thus, a series of investigations has supported
proved difficult to answer. Some individual characteristics, such the proposed linkage between personality, social behavior, and
as self-esteem, internality-externality, and extraversion have measures of satisfaction and well-being.
been found to display consistent associations with measures of A generally similar approach is the recently proposed notion
subjective well-being (see Diener, 1984, for a summary of these of relational competence (Hansson, Jones, & Carpenter, 1984).
efforts), but other characteristics, like intelligence, that might Relational competence refers to those characteristics of the in-
be expected to display such associations have not. Even the con- dividual that "facilitate the acquisition, development, and
sistent associations found in many studies leave important is- maintenance of mutually satisfying relationships" (Hansson et
sues, such as causal direction, unresolved (Diener, 1984). al., p. 273). In turn, this ability to acquire and maintain social
One promising approach to this question is to examine the relationships is assumed to benefit the individual because it al-
role that specific personality factors play in affecting mediating lows greater access to potentially important networks of social
variables, which in turn influence well-being. In particular, re- support. Thus, personality variables are again viewed as influ-
cent work in which important social behaviors served as medi- ences on an intervening variable, the availability of social sup-
ating variables has been successful. For example, in a pair of port networks, which has been demonstrated to influence well-
investigations of high school students, dispositional private self- being (e.g., Cobb, 1976; Dean & Lin, 1977). Of particular im-
consciousness was consistently associated with more intimate portance for our investigation is the fact that Hansson et al.
self-disclosure to peers, which was in turn associated with less (1984) distinguished between the dispositional characteristics
subjective loneliness (Davis & Franzoi, 1986; Franzoi & Davis, most important in constructing new relationships or accessing
1985). This link between private self-consciousness and self-dis- available networks (e.g., shyness, assertiveness) and those char-
closure was replicated with a sample of college-age dating cou- acteristics more useful in maintaining the quality of existing
ples; in addition, disclosure to dating partner was found to pre- relationships (e.g., empathy, emotional stability). It is the latter
dict another kind of well-being: satisfaction with the dating rela- type of competence with which this investigation is concerned.
tionship (Franzoi, Davis, & Young, 1985), More recently, two
aspects of dispositional empathy (empathic concern and per- Dispositional Empathy and Relational Competence
sonal distress) were found among high school students to be reli-
ably associated with such variables as number of friends, num- Hansson et al. (1984) and Hansson (1985) have identified em-
ber of dates, and degree of self-disclosure; these variables were pathy as a characteristic especially likely to influence the main-
tenance of existing relationships. Given the general nature of
empathy—one person's responsivity to the experiences of an-
other—this seems a reasonable conclusion. It seems highly
We gratefully acknowledge the useful comments of Linda A. Kraus
and Stephen L. Franzoi on an earlier version of this article. likely that one's responses to others, both cognitive and affec-
Correspondence concerning this article should be addressed to Mark tive, will play a significant role in enhancing or diminishing
H. Davis, who is now at Behavioral Science, Eckerd College, St. Peters- one's relational competence.
burg, Florida 33733. Rather than treat dispositional empathy as a unidimensional

397
 398 MARK H. DAVIS AND H. ALAN OATHOUT

construct, we view it multidimensionally (Davis, 1980, 1983c; The nature of this model allows several kinds of questions to
Deutsch & Madle, 1975) and measure it accordingly. Three fac- be posed. We address each in turn.
ets of empathy are considered, all of them assessed via the Inter- What impact will these separate facets of empathy have on
personal Reactivity Index (Davis, 1980). These three disposi- interpersonal behavior? On the basis of previous research using
tions are perspective taking, the cognitive tendency to see things this scale (e.g., Davis, 1983c), we strongly expect that perspec-
from another's point of view; empathic concern, the tendency tive taking and empathic concern will be associated with more
to experience feelings of sympathy and compassion for others; frequent positive behaviors and less frequent negative behav-
and personal distress, the tendency to experience personal feel- iors, whereas personal distress will display the opposite pattern.
ings of distress and anxiety in the presence of distressed others.' What impact will specific behaviors have on partner satisfac-
Thus, the Perspective Taking scale taps a cognitive facet of em- tion? Although perceptions of all the positive social behaviors
pathy, whereas the Empathic Concern and Personal Distress are expected to display positive associations with satisfaction,
scales tap different types of affective responsivity. Evidence con- and perceptions of all the negative behaviors are expected to
cerning the validity of these scales comes from numerous recent display negative associations, it is not clear which perceptions
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

investigations (Davis, 1980, 1983a, 1983b, 1983c; Davis, Hull, will emerge as most important in influencing satisfaction. In
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Young, & Warren, 1987; Franzoi et al., 1985). particular, when a path-analytic approach is taken, multicoUin-
The social arena within which we chose to study empathy's earity among these variables will tend to render some of them
impact on relational competence was the romantic relation- nonsignificant as predictors of relationship satisfaction. A re-
ship. Because such relationships are important social arrange- lated issue that can be addressed in this investigation is the rela-
ments providing considerable social support, romantic partners tive importance of positive and negative perceptions as a class
are typically motivated to maintain the quality of the relation- in affecting satisfaction. As Brehm (1985, p. 155) has noted, the
ships as successfully as possible. The supposition guiding this evidence is contradictory regarding the relative importance of
research is that personality in general and empathy in particular costs and benefits in nondistressed intimate relationships. Some -
facilitate relationship quality by affecting the likelihood that research has suggested that the pleasing behaviors of partners
specific social behaviors will occur within the context of that generally influence satisfaction more strongly than do negative
relationship. Some of these behaviors (e.g., affection) will tend behaviors (e.g., Jacobson, Waldron, & Moore, 1980; Rusbult,
to strengthen the relationship, and others (e.g., insensitivity) 1983), whereas others have found negative partner behaviors
will tend to weaken it. Although this does not rule out other generally more influential (e.g., Wills, Weiss, & Patterson,
potential influences of personality on well-being, the emphasis 1974). The design of this investigation allows a direct compari-
in this model on the mediating role of social behaviors seems son of these two classes of interpersonal behavior.
especially useful because it holds the promise of identifying spe- What differences exist between men and women? It is possi-
cifically how personality is translated into well-being. ble to evaluate with this model whether the hypothesized medi-
ating process is comparable for men and women; that is, does
male personality ultimately influence female satisfaction in the
The Model
same ways and to the same extent as female personality affects
Figure I displays the general model used in this investigation. male satisfaction? One piece of evidence bearing on this ques-
far the most part, it reflects the discussion of the issues thus far. tion comes from the Franzoi et al. (1985) investigation, which
However, two other points shoud be stressed. suggests that female personality more broadly influences male
First, the model predicts that the empathic predispositions of satisfaction than vice versa.
one partner will eventually lead to increased satisfaction for the Does the model work in the same way early in the relationship
other partner: Male perspective taking, for instance, will influ- as it does later? The possibility exists that some aspects of this
ence male behaviors that eventually result in increased or de- model will work more or less strongly depending on the length
creased satisfaction for the female partner. This model, there- of the relationship. For example, are some kinds of interper-
fore, assumes that a key element in the successful maintenance sonal behavior more influential in affecting satisfaction early in
of a romantic relationship is to act in a way that renders one's a romantic relationship but less important later? Although this
partner happy with the arrangement. There are, no doubt, lim- is not a longitudinal investigation, it is possible to explore this
its to this; it is possible to conceive of people devoting them- issue by separately testing the model for those couples who have
selves so successfully to satisfying their partners that their own been together for shorter and longer periods of time.
satisfaction with the relationship suffers as a result. However,
within the normal range of relationship activity, it seems likely Method
that behaviors that increase the partner's liking for the relation-
ship will strengthen that relationship and thus benefit the indi- Subjects
vidual in question. The sample consisted of 264 heterosexual student couples at Eastern
Second, the model reflects a phenomenological approach in Illinois University. Of these couples, 29 (! t %) were married or c
that it assumes that satisfaction with the relationship is directly
influenced not by one's partner's behavior but by perceptions 1
The fourth Interpersonal Reactivity Index scale, the Fantasy scale,
of the partner's behavior. The model thus predicts a significant was not included in these analyses. This scale measures the respondent's
connection between self-reported behavior and one's partner's tendency to put himself or herself into the place of fictitious characters
perception of that behavior but identifies that perception as the in books, movies, and plays and did not seem as relevant to the real-tife
more powerful influence on the partner's resulting satisfaction. relationship behaviors as the other three scales.
 MAINTENANCE OF ROMANTIC RELATIONSHIPS 399

EMPATHIC
CONCERN

PARTNER'S
SPECIFIC PERCEPTIONS
PERSPECTIVE^ PARTNER'S
-»• RELATIONSHIP- ->-OF SPECIF1C -
TAKING BEHAVIORS SATISFACTION
RELATIONSHIP
BEHAVIORS
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

PERSONAL/'
DISTRESS
Figure 1. General theoretical model.

the rest reported that they were dating one another exclusively. Approxi- Davis, 1982; Davis, 1983a; 1983b; 1983c; Davis etal., 1987) have pro-
mately 33% of the subjects were 18 years of age or younger, 17% were vided evidence that each scale indeed measures the quality intended.
age 19, 18% were age 20, 12% were age 21, and 20% were 22 or older. Degree of satisfaction with the relationship was assessed by using four
Roughly one third of the couples (77) had been together 6 months or slightly modified items from the Marital Adjustment Test (Locke,
less; 35 couples had been together 7-12 months, 38 couples 13-18 1951). Each item was modified so that it conformed to a standard 5-
months, 23 couples 19-24 months, and 76 couples for over 24 months. response Likert scale and could apply to either married or unmarried
couples. These four items asked directly about degree of relationship
happiness ("How frequently do you and your partner get on each other's
Procedure nerves?", "Have you ever wished you were not going with your part-
ner?", "How happy would you rate your relationship?", and "How satis-
Subjects were recruited from psychology classes and qualified for the fying to you is your relationship with your partner?"); responses could
study if they were currently involved in an intimate and monogamous range from almost always/very happy/very unsatisfying to almost never/
heterosexual relationship. Because in most cases both members of the very happy/very satisfying. Responses to these four items were summed
relationship were not enrolled in the target classes, it was necessary to (reversing the scoring of the two negative items), then divided by 4 to
have those who were enrolled enlist their partners in the study. Enrolled produce a single measure of satisfaction.
subjects completed and returned the Relationship Questionnaire dur- The final set of questions ascertained the frequency with which 25
ing a 25-min period during the normal class hour and were given a blank specific interpersonal behaviors were displayed by each partner. These
Relationship Questionnaire for their partners to complete in private. questions appear in the Appendix.1 Each participant was required to
Subjects who completed the questionnaire in class placed their Social rate both self and partner on each behavior. Each rating was carried out
Security number on both forms so that proper identification could be by use of a 5-point Likert-type response scale running from 1 (J/my
made once their partners had responded to the survey. On receipt of
the nonenrolled subjects' completed questionnaires (returned through
campus mail or by nonenrolled subjects1 partners), couples' responses 1
Rather then decide a priori which behaviors to include, we instead
were matched and tabulated. enlisted the aid of an independent sample of college undergraduates for
this decision. In the semester preceding this study, a group of 100 under-
Relationship Questionnaires graduates at Eastern Illinois University were asked to respond to two
questions about the behavior of their romantic partners. First, they were
Subjects were asked to provide information concerning their sex, age, asked to list as many behaviors of their partner as they could that
marital status, and length of the relationship. The three personality vari- "bother you, irritate you, displease you, or make you unhappy in some
ables included in the questionnaire were three measures of disposi- way." Second, they were asked to list as many behaviors of their partner
tional empathy: perspective taking, empathic concern, and personal dis- as they could that "make you happy, or comfortable, or contented and
tress. These variables were assessed by 7-item subscales from the Inter- satisfied in your relationship." The 100 respondents produced over 1000
personal Reactivity Index (IRI; Davis, 1980). The scales—the Perspec- answers (696 positive and 437 negative). By eliminating the duplicates
tive Taking (PT), Empathic Concern (EC), and Personal Distress (PD) and excluding unusual or highly infrequent responses, we reduced this
scales—all possess adequate internal reliability (a = .73, .71, and .76, list to a set of 25 behaviors; these are the items that appear in the Ap-
respectively; Davis, 1980). Several investigations (e.g., Bernstein & pendix.
 400 MARK H. DAVIS AND H. ALAN OATHOUT

partner never acl(s) this way) to 5 (I/my partner very often actfs) Regarding the first step in the model, empathic concern ex-
this way). erted the most widespread effect, with a total of 8 significant
For convenience, both statistical and conceptual, the number of be- paths leading from empathic concern to self-reported behavior
havior items was further reduced before analysis was begun. On the
(3 paths for men; 5 for women). Perspective taking displayed 7
basis of the conceptual similarity of some items, smaller sets of items
significant paths (2 for men; 5 for women), whereas personal
were combined to produce a set of seven behavior clusters (four positive
distress displayed 6 such paths (2 for men; 4 for women). Impor-
and three negative). For each cluster, the items comprising it were
summed, then divided by the number of items making up that cluster; tantly, the sign of every one of these 21 significant paths was in
this procedure was carried out separately for each respondent's seven the predicted direction; that is, empathic concern and perspec-
self-rating clusters and his or her seven partner-rating clusters. The clus- tive taking were always positively associated with positive self-
ters, and the items making up each one, are as follows: The three nega- reported behaviors and negatively associated with negative self-
tive clusters are insensitivity (made up of the items rude, critical/nag- reported behaviors. Personal distress consistently displayed the
ging, dominating, selfish/egocentric, immature, and unpleasant habits), opposite pattern: negative associations with positive self-re-
untrustworthiness (made up of items untrustworthy anA excessive atten- ported behaviors and positive associations with negative ones.
tion to opposite sex), and possessiveness (made up of the single item
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

In terms of the sheer number of significant paths, it is clear

possessive). The positive clusters are warmth (made up of affectionate/
This document is copyrighted by the American Psychological Association or one of its allied publishers.

that the empathic predisposition of women had stronger effects

loving, supportive, thoughtful/considerate, and generous/giving), even
on their self-reported behavior than the empathic predisposi-
temper (made up of even tempered'and patient/understanding), positive
outlook (made up of confident, friendly/outgoing, positive/optimistic, tion of men had on their self-reported behavior (14 significant
ambitious, honest/dependable, and intelligent), and good communica- paths of 21 possible for women; 7 of 21 for men). The difference
tion (made up of opens up to partner, readily listens to partner, does not between men's and women's personality effects can also be seen
open up to partner, and does not listen to partner. )3 in a comparison of the variance accounted for in the self-re-
ported behavior of each sex. For men, the mean S2 for their
seven self-reported behaviors is .08; for women the mean /J2 is
Results
.14. Thus, the initial stage in the model seems to fit the data
Sex Differences better for female than for male respondents.5

Table 1 displays separately for men and women the mean 3

Although these combinations of items were derived on conceptual
scores on all the major variables in this investigation. rather than empirical grounds, they appear, statistically speaking, to be
Consistent with other reported sex differences involving the coherent groupings. For those clusters consisting of at least three behav-
IRI (e.g., Davis, 1980), female respondents scored significantly ior items, the alpha coefficients are as follows: For good communication,
higher on all three of the empathy scales. Although men and alphas ranged from .61 to .77 for the four indexes created by the 2 (male,
women reported statistically equivalent levels of satisfaction female) X 2 (self, partner) matrix of ratings; for warmth, .69 to .76; for
with the relationship, they differed significantly on a number of positive outlook, .66 to .73; and for insensitivity, .69 to .76. The mean
the ratings of behavioral clusters. For example, when making alpha across all of these clusters is .70. The even-temper and untrust-
worthiness clusters consist of only two items each, and alpha coefficients
self-ratings, women reported significantly higher levels of posi-
cannot be calculated; the average interitem correlation for those clusters
tive behaviors and significantly lower levels of negative behav-
are .22 and .40, respectively. Thus, these item clusters indeed comprise
iors than did men for three of the seven behavioral clusters.4
coherent groupings. Also, two of the items making up the good-commu-
Although this may indicate that women, relative to men, more nication cluster actually ask about poor communication behaviors {does
frequently act in positive ways and less frequently act in negative not open up to partner and does not listen to partner). These two items
ways, it seems more likely that it reflects a generally more posi- were reverse scored and added to the good-communication cluster be-
tive response set among female respondents. When both men cause they were so highly correlated with the two positively worded
and women rate the male partner, for example, women again items. (Mean correlation of positively and negatively worded items was
make more favorable ratings in four of seven opportunities. The .41 for self-reports and .57 for judgments of one's partner.) Given this
same pattern holds generally true for ratings of the female part- high correlation, it seemed more appropriate to combine them into one
measure.
ner (two of seven) and ratings of partner (two of seven). 4
Because of the high number of nonindependent i-test comparisons
reported here, only comparisons reaching the more stringent probabil-
Evaluating the Model ity level of .001 are considered significant.
1
Although the theoretical model in Figure I holds that empathic pre-
The model outlined earlier was evaluated by means of path dispositions lead to behavior, only self-reports of actual behavior are
analysis (Pedhazur, 1982; Wright, 1934). The criteria used to assessed in this investigation. Such self-reports unquestionably bear
construct the path model were as follows: some relationship to actual behavior, but admittedly it is not a perfect
1. Paths were estimated from all three empathy measures of one. Thus, the most convincing evidence for this model will eventually
an individual to all seven of his or her self-reported behaviors. result from investigations that involve measures of actual behavior,
such an assessment is unfortunately beyond the scope of this study.
2. One path was estimated from each self-reported behavior
However, for at least two reasons, the results involving self-reported be-
of the individual to the corresponding behavior as perceived by
havior in this study are meaningful. First, the IRI has repeatedly been
the partner. found to be associated with behavioral variables other than simple self-
3. Paths were estimated from each of the partner's percep- reports (e.g., Bernstein & Davis, 1982; Davis, 1983a; Neale & Bazer-
tions of behavior to the partner's satisfaction. nian, 1983). Second, the consistently significant association found in
The results of these path analyses are displayed separately for this investigation between self-reported behavior and partner's percep-
men and women in Figures 2 and 3. tions of the behavior further increases our confidence that self-reports
 MAINTENANCE OF ROMANTIC RELATIONSHIPS 401

Table 1 live power is substantial. For both sexes, satisfaction was sig-
Mean Differences Between Sexes on Major Variables nificantly influenced by perceptions of both positive and nega-
tive partner behaviors: Satisfaction was diminished by percep-
Variable Men Women
tions of partner possessiveness and Untrustworthiness and
Personality variables increased by perceptions of partner warmth and positive out-
Perspective taking 15.9 17.5' look. In addition, for women only, a perception of good commu-
Empathic concern 19.1 22.3* nication with her partner also enhanced satisfaction.
Personal distress 9.7 13.0* Because one goal of this investigation was to examine the rel-
Satisfaction 16.4 16.6
Self-reported behavior ative importance of positive and negative behaviors, we carried
Good communication 3.7 4.1* out an additional set of analyses. In these analyses, male and
Warmth 4.0 4.3* female satisfaction each served as dependent variables in two
Even temper 3.7 3.7 hierarchical regression equations. In the first step of the first
Positive outlook 4.0 4.1
equation, perceptions of the four positive behaviors were en-
Untrustworthiness 2.2 2.0
tered as predictors, and on the second step, perceptions of the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Insensitivity 2.6 2.3*

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Possess iveness 2.6 2.7 three negative behaviors were entered. This procedure makes it
Ratings of male behavior possible to ascertain whether the addition of the negative behav-
Good communication 3.7 3.9 iors significantly increases the equation's predictive power be-
Warmth 4-0 4.2*
3.7 yond that already provided by the positive behaviors. Then we
Even temper 3.9
Positive outlook 4.0 4.1* repeated the procedure, this time entering perceptions of the
Untrustworthiness 2.2 2.0' negative behaviors first, followed by perceptions of the positive
Insensitivity 2.6 2.4* behaviors.
Possessiveness 2.6 2.6
We found that each set of behavioral perceptions did indeed
Ratings of female behavior
Good communication 3.9 4.1* make a unique significant contribution to the predictive power
Warmth 4.2 4.3* of the equation. For male satisfaction, the inclusion of the nega-
Even temper 3.6 3.7 tive behaviors significantly increased the/f2 above that obtained
Positive outlook 4.0 4.1* by using only the positive behaviors (R* increase from .26 to
Untrustworfhiness 2.0 2.0
.34), F(4, 329) = 6.86, p < .01. Entering the negative behaviors
fnsensitivity 2.2 2.3
Possessiveness 2.8 2.7 first and then the positive behaviors also produced a significant
Ratings of partner behavior improvement in R2 (from .21 to .34), F(3, 239) = 15.32, p <
Good communication 3.9 3.9 .01. Because of the size of the R2 increase, however, it appears
Warmth 4.2 4.2
that perceptions of positive behaviors are exerting the greater
Even temper 3.6 3.9*
4.0 4.1* unique effect on satisfaction.
Positive outlook
Untrustworthiness 2.0 2.0 The same general pattern was found for female satisfaction.
Insensitivity 2.2 2.4 Including the negative behaviors significantly increased the R2
Possessiveness 2.8 2.6 above that obtained by using only the positive behaviors (R2
increases from .31 to .36), F(4, 242) = 4,62, p < .01. Entering
Note, For all comparisons. iVs range from 255 to 264 (A's vary due to
missing data). the negative behaviors first and then the positive behaviors also
*p<.001. produced a significant improvement in R2 (from .20 to .36),
f\3, 242) = 21.23, p < .01. Again, it is clear that perceptions of
positive and negative partner behavior make unique contribu-
tions to female satisfaction, although the impact of positive be-
Evidence was strong for the second step of the model: the link
haviors appears greater.6
between self-reported behavior and partner's perception of the
behavior. For both men and women, significant positive paths
emerged for each of the seven behavioral clusters. These paths Length of Relationship
ranged in size from .27 to .47, with a mean of .39 for women's A final question concerned the possibility that trails, percep-
perceptions and .35 for men's perceptions. tions, or behaviors important early in a romantic association
Turning to the final step in the model, it is apparent that the would be less important later or vice versa. To explore this no-
model's ability to account for variation in satisfaction is consid-
erable. For each sex, approximately one third of the variation
in relationship satisfaction is accounted for by one's percep- 6
One caution should be voiced concerning such comparisons of vari-
tions of partner behavior {R2 = .36 for female satisfaction and ance accounted for C/?2). Although such comparisons can be quite use-
R2 = .34 for male satisfaction). Given the numerous other deter- ful, different operationalizations of the same variable CAB produce quite
minants of such satisfaction not included in the model (e.g., different estimates of R2. As a result, the variances accounted for is this
physical attractiveness; shared values and interests), this predic- investigation may be quite different from those accounted for in other
studies using different methodologies. The If comparisons in this arti-
cle are included because they provide a useful, but by no means perfect,
comparison of the relative predictive power of certain variables or
are indeed tapping the actual observable behavior of the respondent. classes of variables. Though informative, they should be viewed with
Nevertheless, this study deals only with reports of behavior. proper caution.
 402 MARK H. DAVIS AND H. ALAN OATHOUT

MALE MALE SELF- FEMALE PERCEPTIONS FEMALE

PERSONALITY REPORTED BEHAVIOR OF MALE BEHAVIOR SATISFACTION

GOOD COMMUNICATION- GOOD COMMUNICATION

(F?- .09)
.22

.44
EMPATHIC WARMTH WARMTH
CONCERN
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

.29
This document is copyrighted by the American Psychological Association or one of its allied publishers.

EVENTEMPER• EVEN TEMPER

(F?-.04)

PERSPECTIVE •37 16 FEMALE

POSITIVE OUTLOOK *• POSmVE OUTLOOK —li2- *" SATISFACTION
TAKING
(F?-.36)

UNTRUSTWORTHINESS *• UNTRUSTWORTHINESS
(F?-.05)

PERSONAL .37
INSENSITIVITY INSENSITIVITY
DISTRESS
(F?= .06)

.37
POSSESSIVENESS POSSESSIVENESS
(F?=.05)
Figure2. Full sample model predicting female satisfaction,

tion, we split the sample into two groups: 117 couples who re- to self-reported behavior in the Ml sample analyses, only 12
ported having been together for 12 months or less and 133 cou- such paths emerged for the short relationships.
ples who reported having been together for longer than 12 Although female empathic predispositions again had more
months. Analyses identical to those already reported were car- impact on self-reported behavior than did male predispositions
ried out on these two groups. The results of the analyses involv- (7 paths vs. 5), the difference was not as strong as with the full
ing the shorter relationships appear in Figures4 and 5. sample. Likewise, the mean R* for the seven self-reported fe-
Shorter relationships. As with the full sample, analyses with male behaviors was greater than that of men (. 10 vs. .07), but
the shorter relationship sample revealed that empathic concern not as great as it had been for the full sample.
displayed a broader impact on self-reported behavior than did Evidence for the second step of the model—the link between
the other empathy measures. The empathic concern scale dis- self-reported behavior and partner's perception of the behav-
played a total of 6 significant paths to self-reported behavior (2 ior—was again strong and comparable with that observed for
for men; 4 for women); perspective taking displayed only 3 paths tiie full sample. These paths ranged in size from .23 to .58, with
(3 for men; 2 for women), as did personal distress (2 for men; 1 an average size of .41 for women's perceptions and. 38 for men's
for women). Also paralleling the full sample analyses, in each perceptions.
case the sign of the path coefficient was in the expected direc- Turning to the final step, a considerable degree of variation
tion. The biggest difference between these analyses and the ear- in satisfaction was explained by the model: an amount compa-
lier ones was the overall sparseness of significant effects. rable in size to that explained for the full sample. The 81 for
Whereas the empathy variables exhibited 21 significant paths female satisfaction in the short relationships was ,33, and for
 MAINTENANCE OF ROMANTIC RELATIONSHIPS 403

FEMALE FEMALE SELF- MALE PERCEPTIONS MALE

PERSONALITY REPORTED BEHAVIOR OF FEMALE BEHAVIOR SATISFACTION

.43
GOOD COMMUNICATION —*> GOOD COMMUNICATION

EMPATHIC -46 WARMTH

.44
WARMTH
CONCERN
(I?-.25)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

.30 ,25
EVENTEMPER- ->• EVEN TEMPER
(F?= .09)

.39 .21 MALE

POSITIVE OUTLOOK ———fr- POSITIVE OUTLOOK SATISFACTION
(F?=.19) (F?-.34)

UNTRUSTWORTHINESS !->• UNTRUSTWORTHINESS

(F?= .08)

PERSONAL
INSENSITIVITY •
.31 INSENSITIVITY
DISTRESS

POSSESSIVENESS- .27 •>> POSSESSIVENESS

(F?-.03)
Figure 3. Full sample model predicting male satisfaction.

men it was even greater (.41). In terms of specific influences on plained variance above that accounted for by positive behav-
satisfaction, female satisfaction was still influenced by percep- iors (R2 increase from .31 to .41), F[4, 109) = 4.48; p < .01,
tions of good communication with her partner and by percep- similarly, the addition of perceived positive behaviors signifi-
tions of his possessiveness. However, the influences of perceived cantly increased the R2 above that explained by negative behav-
warmth, positive outlook, and untrustworthiness were no iors (from .28 to .41), /=tt 109) = 8.31, p< .01.
longer evident. For male satisfaction, there was little difference With regard to female satisfaction, however, this pattern did
between the full sample and the short-relationship analyses; not hold. The addition of perceived positive behaviors did sig-
male satisfaction was still significantly influenced by percep- nificantly increase the R2 (from .14 to .33), F{4, 109) = 7.64,
tions of positive outlook, untrustworthiness, and possessive- p < .01, but the addition of perceived negative behaviors did
ness. The only influence not apparent at the early stage of the not significantly improve the predictive power afforded solely
relationship was the effect of warmth. Thus, for both men and by positive behaviors (R2 increase from .28 to .33), F[3,109) =
women, perceived warmth of the partner played no significant 2.15, ns. Thus, although one perception of negative behavior
role in affecting satisfaction for the shorter relationships. (possessiveness) had a significant impact on female satisfaction,
To compare the relative impact of perceived positive and neg- perceived negative behaviors as a class did not significantly im-
ative behaviors on satisfaction, hierarchical regression analyses prove the predictability of female satisfaction.
identical to those discussed earlier were carried out with this Longer relationships. The results of the analyses involving
Short-relationship sample. For male satisfaction, the addition of the longer relationships are displayed in Figures 6 and 7. In con-
perceived negative behaviors significantly increased the ex- trast to the earlier results in which empathic concern displayed
 404 MARK H. DAVIS AND H. ALAN OATHOUT
MALE MALE SELF- FEMALE PERCEPTIONS FEMALE
PERSONALITY REPORTED BEHAVIOR OF MALE BEHAVIOR SATISFACTION

GOOD COMMUNICATION • GOOD COWMUMCATION

(E?-.06)

EMPATH .43
WARMTH -». WARMTH
CONCERN
(F? = .10)
.30
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

.29
This document is copyrighted by the American Psychological Association or one of its allied publishers.

EVEN TEMPER EVEN TEMPER

(F? = .04)

PERSPEC
.41 FEMALE
->• POSmVEOUTLOOK SATISFACTION
TAKING
(F?= .33)

UNTRUSTWORTHINESS UNTRUSTWORTHINESS
(F?- .06)

'-.22

PERSONAL .42
INSENSITIVITY INSENSITIVITY
DISTRESS
(F?- .05)
.22

.41
POSSESSIVENESS POSSESSIVENESS
(F?- .09)
Figure 4 Predicting female satisfaction in shorter relationships.

the broadest impact on self-reported behavior, here perspective self-reported behaviors, the mean R1 was a sizeable .19 (compared
taking had the most consistent effect, displaying 8 significant with. 10 for the short-relationship sample and .13 for the full sam-
paths (3 for men and 5 for women). Empathic concern displayed ple). For those persons involved in longer relationships, then, a
6 significant paths (3 for men; 3 for women) and persona! dis- greater proportion erf their self-reported interpersonal behavior is
tress 5 significant paths (3 for men; 2 for women). Thus, com- accounted for by empathic predispositions.
pared with the shorter term relationships, perspective taking Evidence for the second step of the model was quite similar
among the couples with the greatest longevity had more than to that already reported. Coefficients between self-reported be-
twice as many significant effects on self-reported behavior. As havior and partner perceptions of the behavior for this sample
with all the path analyses thus far, the signs of all of the signifi- ranged from .24 to .46, with an average size of .38 for female
cant paths involving empathic predispositions were in the pre- perceptions and .33 for male perceptions.
dicted direction. Finally, the total number of significant paths Turning to the final step of the model, roughly one third or
involving empathy was much closer to that found for the Ml more of the variation in satisfaction was again accounted for by
sample (full sample, 21 paths; long relationships, 19; short rela- the mode!. The R* for female satisfaction is .42 and for male
tionships, 12). satisfaction is .32. In terms of specific influences on male satis*
Furthermore, the size of the paths from empathy to self-re- faction, the same four perceptions of partner behavior that are
ported behavior was generally stronger, and this is reflected in the significantly related to male satisfaction for the full sample—
average size of the Rz for the self-reported behaviors. For male be- warmth, positive outlook, untrustworthmess, and possessive-
haviors, the average R1 for the longer relationship sample was. 11 ness—were the four significant predictors of male satisfaction
(compared with .07 for both the short and full samples); for female among the longer term sample.
 MAINTENANCE OF ROMANTIC RELATIONSHIPS 405

FEMALE FEMALE SELF- MALE PERCEPTIONS MALE

PERSONALITY REPORTED BEHAVIOR OF FEMALE BEHAVIOR SATISFACTION

GOOD COMMUNICATION- GOOD COMMUNICATION

EMPATHIC .46
WARMTH WARMTH
CONCERN
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

.23
EVEN TEMPER ->• EVEN TEMPER
(F?= .08)

.39 .21 MALE

POSmVE OUTLOOK POSmVE OUTLOOK SATISFACTION
(F?=.41)

UNTRUSTWORTHINESS - UNTRUSTWORTHINESS
(F?-.06)

.37
INSENSITIVITY INSENSITIVITY
DISTRESS
(^=.09)

.25
POSSESSIVENESS POSSESSIVENESS

FigureS. Predicting male satisfaction in shorter relationships.

For female satisfaction, the picture was not as simple. For the ior significantly increased the satisfaction № (from .37 to .42),
longer relationships, perceptions of three positive behaviors— F(4,125) = 2.65, p < .05, as did the inclusion of perceived posi-
2
good communication, warmth, and positive outlook—that tive behavior (R increase from .26 to .42), F(3, 125) « 11.40,
were significant for the full sample were also significant for this p < .01. Thus, unlike in the shorter relationships, female satis-
group. None of the perceptions of negative behavior displayed faction in longer relationships was significantly and uniquely
a significant path to female satisfaction for the longer term cou- influenced by perceptions of negative partner behavior. How-
pfes. The one perception that consistently affected female satis- ever, as the size of the R2 increase indicates, the contribution
faction for both short and long relationships was perceived good of perceived negative behaviors to female satisfaction appears
communication with her partner. Perceived possessiveness, noticeably weaker than the contribution of positive behavior.
which influenced satisfaction among the women in shorter rela-
tionships, failed to do so for those in the longer relationships. Discussion
Analyses to test the relative power of positive and negative
behaviors again revealed the pattern found for the full sample. As a whole, the results of this study ofife clear-cut support
For men, the inclusion of perceived negative behaviors signifi- for the theoretical model. Three separate facets of dispositions!
2
cantly increased the explained variation in satisfaction (R in- empathy had significant, predictable, and discriminable effects
crease from .23 to .32), f\4, 122) - 3.84, p < .01, as did the on self-reports of important social behaviors; these self-reports
inclusion of perceived positive behaviors beyond that of nega- were significantly related to a romantic partner's perceptions of
2
tive behaviors (R increase from .19 to 32), F(3,122) = 7.36, the behaviors; and these partner perceptions in turn exerted a
p < .01. For women, the inclusion of perceived negative behav- considerable impact on the partner's satisfaction with the rela-
 406 MARK. H. DAVIS AND H. ALAN OATHOUT

tionship, accounting for roughly one third of the total variation tive power. That is, more powerful emotional responses may be
in such satisfaction. Thus, the general tenets of the model— typically evoked by acts of jealousy and suspicion or by acts of
that personality can influence one's relational competence in warmth and affection than are produced by the less dramatic
romantic relationships through its effect on mediating behav- irritations resulting from partner insensitivity or the more
iors—was strongly supported. mundane satisfaction resulting from a patient and understand-
ing partner. Perceived partner insensitivity and even temper
What Impact Did These Separate Facets of Empathy might therefore pale in comparison as predictors of overall sat-
isfaction.
Have on Behavior?
Another way to conceptualize this explanation is in terms of
For every analysis, using full or partial samples, each signifi- the cognitive processes that underlie social inferences. For ex-
cant path from empathy to behavior held the expected sign. Per- ample, when asked to estimate one's satisfaction with a rela-
spective taking and empathic concern were always associated tionship, an individual may be especially likely to retrieve and
with more self-reports of positive behaviors and fewer of nega- consider some instances of partner behavior more readily than
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tive behaviors, whereas personal distress showed the opposite others. More vivid or extreme behaviors (such as those involv-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

pattern. These findings validate the decision to use a multidi- ing issues of jealousy and suspicion) may be especially likely to
mensional conception of empathy; in this study, the impact of be recalled, whereas less vivid behaviors may be underretrieved
an empathic predisposition clearly depends on the precise na- (Fiske & Taylor, 1984, p. 250). As a result, satisfaction judg-
ture of that disposition, and a global measure of empathy would ments may be more influenced by perceptions of dramatic,
not have revealed this. memorable partner behaviors and less influenced by less mem-
Which aspect of empathy plays the largest role? One way to orable ones.
address this question is simply to consider the number of self- The second approach to studying the impact of perceptions
reported behaviors influenced by each facet of empathy. This on satisfaction is to consider the influence of positive and nega-
approach shows that the three individual difference measures tive behaviors as a class. This approach revealed that for both
are roughly equivalent, with empathic concern (eight significant men and women, satisfaction was significantly and uniquely
paths, including both men and women) having a slightly affected by each class of behavior. However, in each case the
broader impact than perspective taking (seven paths), which in class of perceived positive behaviors seemed to have a larger
turn has a slightly broader effect than personal distress (six unique effect on satisfaction than did the class of perceived neg-
paths). Because not all of the self-reported behaviors eventually ative behaviors. This effect is particularly pronounced for fe-
lead to partner satisfaction (e.g., even temper), a second means male satisfaction.
by which to address this question is to ask how many self-re- This pattern is generally consistent with the findings of Jacob-
ported behaviors that ultimately influence partner satisfaction son et al. (1980) and Rusbult (1983), which also revealed a gen-
are affected by each empathic disposition. By this criterion, em- erally stronger effect of positive partner behavior on satisfaction
pathic concern still remains most influential (six paths), fol- among nondistressed romantic couples. In the Rusbult (1983)
lowed by personal distress (five paths) and perspective taking investigation, perceived costs in the relationship (negative be-
(two paths). Thus, generally speaking, empathic concern seems havior) had no significant overall effect on satisfaction for men
to play a more powerful role in this model. or women; in the Jacobson et al. (1980) study, as here, both
positive and negative behaviors influenced satisfaction for each
What Impact Did Behavioral Perceptions Have on sex, but negative behaviors did so less strongly. Similarly,
though positive behavior does seem to play a larger role in this
Partner Satisfaction?
study, both classes of behavior did have unique influences on
There are two approaches to addressing this question. The satisfaction; neither class of partner behavior alone was suffi-
first is to consider the impact of partner's perceptions of specific cient to predict satisfaction optimally.
behaviors on partner's satisfaction. From this perspective, the
pattern is very clear: Both male and female satisfaction are in- What Differences Exist Between Men and Women?
fluenced by the same four perceptions of partner behavior. Per-
ceptions of partner warmth and positive outlook enhance one's Several interesting differences between men and women were
satisfaction with the relationship, whereas perceptions of part- evident in these results. The first of these concerns the impact
ner possessiveness and untrustworthiness diminish satisfaction. of personality variables on self-reported behavior. As Figures 2
In addition, for women only, satisfaction is significantly influ- and 3 indicate, the empathy scores of women were significantly
enced by perceptions of good communication; moreover, this related to twice as many self-reported behaviors (14) as male
path coefficient is the largest of the five significant predictors of personality scores were to male behaviors (7). This pattern is
female satisfaction. mirrored by the R2 values associated with male and female self-
It is difficult to offer a fully convincing post hoc explanation reported behavior. In almost every instance, the amount of vari-
for the finding that perceived partner insensitivity and even ation in self-reported female behavior accounted for by female
temper consistently fail to predict satisfaction. One possible ex- personality was greater than the comparable variance ac-
planation for this phenomenon rests on the argument that in- counted for in men. The mean R2 for female self-reported be-
sensitivity and even temper are in some sense less dramatic be- havior was .13, and for men it was only .07. Thus, the impor-
haviors than are those involving issues of trust and possessive- tance of personality in this model was noticeably stronger for
ness and that this may account for their relative lack of predic- the women.
 MAINTENANCE OF ROMANTIC RELATIONSHIPS 407

MALE MALE SELF- FEMALE PERCEPTIONS FEMALE

PERSONALITY REPORTED BEHAVIOR OF MALE BEHAVIOR SATISFACTION

GOOD COMMUNICATION - • GOOD COMMUNICATION

(F?- .13)
.2C

.46 -». WARMTH

WARMTH
(#=.19)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

.29
This document is copyrighted by the American Psychological Association or one of its allied publishers.

EVEN TEMPER EVEN TEMPER

(F?-.05)

FEMALE
PERSPECTIVE POSmVEOUTLOOK — * POSmVEOUTLOOK — SATISFACTION
TAKING
(F?=.42)

UNTRUSTWORTHNESS ' 45 > UNTRUSTWORTHNESS

(F?-.06)

PERSONAL .31 INSENSITIVITY

INSENSITIVITY
DISTRESS
(#=.09)

.34 ->• POSSESSIVENESS

POSSESSIVENESS
(F?- .06)
Figure 6. Predicting female satisfaction in longer relationships.

Franzoi et at. (1985), in a similar investigation of personality faction: good communication. Male satisfaction was unaffected
and relationship satisfaction, also found that women's perspec- by the perception that his female partner was a receptive listener
tive taking exerted a greater effect on male satisfaction than or a willing discloser, in contrast, female satisfaction was more
male perspective taking did on female satisfaction. Their tenta- strongly influenced by this perception than by any other. This
tive explanation was that perspective taking is a characteristic is consistent with previous research that suggests a greater im-
that allows women to carry out the relationship role tradition- pact of disclosure patterns on subjective feelings (like loneli-
ally expected of them, as the partner especially concerned with ness) among women than men (Berg & Peplau, 1982; Solano,
emotional and social needs in the relationship. Thus, perspec- Batten, & Parish, 1982).
tive taking is a role-relevant characteristic for women and may A final intriguing sex difference emerged regarding the im-
be more predictive of partner satisfaction than is male perspec- pact of perceived positive and negative behaviors on satisfac-
tive taking, as the traditional male role does not so clearly em- tion. As mentioned earlier, the unique explanatory power of
phasize this function. The same argument may be applied to positive behaviors is greater than that of negative behaviors, but
our study: Empathy in all its forms may play a more significant this is especially true among women. For men, the unique con-
role in affecting important relationship behaviors for women tribution of perceived positive behavior to satisfaction is .13,
because empathy is usually a more role-relevant characteristic and for negative behavior it is .08. For women, the figures are
for women than for men. . 16 for perceived positive behavior and only .05 for negative be-
Sex differences also emerged concerning the impact of part- havior. The relative power of positive to negative behavior is
ner perceptions on satisfaction. In particular, one perception thus roughly twice as large for women. This pattern was also
emerged as a significant predictor of female but not male satis- found by Jacobson et al. (1980). In that study, 69% of the ex-
 408 MARK H. DAVIS AND H. ALAN OATHOUT

FEMALE FEMALE SELF- MALE PERCEPTIONS MALE

PERSONALITY REPORTED BEHAVIOR OF FEMALE BEHAVIOR SATISFACTION

.35
GOOOCOMMUNCADON GOOD COMMUNICATION
(#=.30)

EMPATHIQ .41
WARMTH WARMTH
CONCERN
(#=.32)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

.37
.31
This document is copyrighted by the American Psychological Association or one of its allied publishers.

EVEN TEMPER EVEN TEMPER

(#=.12)

PERSPEC 25
MALE
** POSmVEOUTLOOK —— POSITIVE OUTLOOK SATISFACTION
TAKING
(#=.24) (#=.32)

UNTRUSTWORTHINESS »- UNTRUSTWORTHINESS
(#= .10)

PERSONAL' .24
INSENSmVITY INSENSITIVITY
DISTRESS
(#=.16)

.28
POSSESSIVENESS •>• POSSESSIVENESS
(#= .06)
Figure 7. Predicting male satisfaction in longer relationships.

plained variation in male satisfaction was due to positive part- counted for by personality variables was greater for longer rela-
ner behavior, whereas 93% of the explained female satisfaction tionships (for the shorter relationships, mean R1 = .07 for men
was accounted for by positive partner behavior. A compelling and. 10 for women; for the longer relationships, mean R2 = . 11
reason for this pattern is difficult to provide, and in the face of and .19, respectively). When only positive self-reported behav-
other studies not reporting it (i.e., Rusbult, 1983; Wills et al., iors are examined, the average R2 for women in longer relation-
1974), it may be wisest not to speculate about one. ships increases to .25.
Why should personality more strongly predict self-reported
Did the Model Work in the Same Way at All Stages of behavior in the long relationships? One explanation rests on the
the Relationship? argument that chronic dispositional tendencies may play a
lesser role in affecting behavior at the early stages of the relation*
Two quite interesting differences emerged between those cou- ship because role constraints, self-presentational concerns, and
ples who had been together for a year or less and those who had other situational forces are more salient then. Because of this,
been together longer. The first of these differences has to do with the opportunity for personality variables to influence behavior
the predictive power of the model: Overall, the model can better are limited. As the relationship progresses, these situational fac-
explain self-reported behavior in longer relationships. For ex- tors recede in importance, and stable personality traits play a
ample, 12 significant paths lead from personality to self-re- commensurately larger role.7
ported behavior for the shorter relationships (for both men and
women) as opposed to 19 for the longer relationships. Likewise, 7
Another possibility is that the self-reports of behaviors in the longer
for virtually every self-reported behavior, the variance ac- relationships are more reliable than those in the shorter relationships.
 MAINTENANCE OF ROMANTIC RELATIONSHIPS 409

The second and perhaps more striking difference between the sexes. Again, however, satisfaction is doubtless dependent on a
findings for the shorter and longer relationships is the much variety of other variables and perceptions.
more consistent effect of perspective taking on self-reported be- A second limitation of the model concerns the sample. Al-
havior among the latter group. Although only three significant though some married couples are included, the majority of re-
paths from perspective taking to self-reported behavior are evi- spondents are student couples involved in steady dating rela-
dent in the shorter relationship sample, eight such paths are tionships. This creates two problems. One is that it becomes
evident in the longer sample. Similarly, none of the perspective- impossible to know if these findings would hold for longer term
taking paths ultimately influence partner satisfaction for the relationships that continue into marriage. The other problem
shorter relationships, whereas four of them do for the longer is that the particular environment in which these relationships
relationships. In contrast, the effect of empathic concern is vir- exist—college life—may also make this sample unrepresenta-
tually the same for the long and short samples, and the influence tive. It is critical to explore this or similar models with largely
of personal distress is only slightly improved (short relation- nonstudent, married samples to answer these questions.
ships, three paths; long relationships, five paths). A third limitation has to do with the comparison of shorter
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

This pattern is consistent with a view of romantic relation- and longer relationships. The implication of these analyses is
This document is copyrighted by the American Psychological Association or one of its allied publishers.

ships as evolving from a relatively intense, emotional beginning that the shorter relationships represent the longer relationships
into a less intense, more companionable union. Perspective tak- at an earlier stage. Although likely, this is not certain. By using
ing, the nonemotional and explicitly cognitive tendency to en- this cross-sectional approach, the possibility exists that the
tertain other points of view, may not be especially important shorter and longer relationship groups differ on more than sim-
as an influence on behavior in the emotional early stages of a ply relationship length. It is possible, for example, that most of
romantic relationship. As time passes and the relationship sta- the shorter relationship couples in this study will not remain
bilizes, a consistent tendency to put aside one's own perspective intact long enough to become longer relationship couples. If so,
and entertain that of the partner begins to pay dividends. Conse- then the real difference between the groups may be unsuccessful
quently, perspective taking becomes a more powerful predictor and successful instead of shorter and longer. Only a longitudinal
of a variety of behaviors that ultimately influence partner satis- design will truly allow an examination of the model at early and
faction. later stages of the relationship.

Caveats References

In addition to its strengths, this investigation has some clear Berg, J. H., & Peplau, L.A. (1982). Loneliness: The relationship of self
limitations. First, the model is by no means an exhaustive one. disclosure and androgyny. Personality and Social Psychology Bulle-
tin, 8, 624-630.
There are certainly influences on behavior other than the three
Bernstein, W. M., & Davis, M. H. (1982). Perspective-taking, self-con-
dispositional empathy measures used here. These three vari-
sciousness, and accuracy in person perception. Basic ami Applied So-
ables were chosen because they seemed especially likely to in-
cial Psychology, 3. 1-19.
fluence one's ability to maintain an existing relationship, and Brehm, S. S. (1985). Intimate relationships. New York: Random House.
the success of the model suggests that this was an appropriate Cobb, S. (1976). Social support as a moderator of life stress. Psychoso-
choice. However, it would be unwise to believe that other dispo- matic Medicine, 38, 300-313.
sitional characteristics could not substantially increase the Davis, M. H. (1980). A multidimensional approach to individual
model's ability to account for self-reported behavior. differences in empathy. JSAS: Catalog of Selected Documents in Psy-
Similarly, there are certainly influences on relationship satis- chology, 10. 85.
faction other than the perceptions of the particular behaviors Davis, M. H. (1983a). The effects of dispositional empathy on emo-
tional reactions and helping: A multidimensional approach. Journal
assessed here. Partner attractiveness and shared values, inter-
of Personality, SI, 167-184.
ests, and attitudes are two factors very likely to influence overall
Davis, M. H. (1983b). Empathic concern and the muscular dystrophy
satisfaction significantly. Given the fact that such variables were
telethon: Empathy as a multidimensional construct. Personality and
not included, it is impressive that the model was able to account Social Psychology Bulletin, 9, 223-229.
for roughly one third of the variation in satisfaction for both Davis, M. H. (1983c). Measuring individual differences in empathy: Ev-
idence for a multidimensional approach. Journal of Personality and
Social Psychology, 44, 113-126.
Davis, M. H., & Franzoi, S. L. (1986). Adolescent loneliness, self-disclo-
This could occur because those self-reports are based on more instances sure, and private self-consciousness: A longitudinal investigation.
of each behavior and thus might produce a more stable assessment of Journal of Personality and Social Psychology, 51, 595-608.
that behavior's frequency. However, an examination of the internal reli- Davis, M. H., Franzoi, S. L., & Wellinger, P. (1985, August). Personality,
ability of the self-report clusters for shorter and longer relationships does social behavior, and loneliness. Paper presented at the annual conven-
not support this view. For the behavior clusters consisting of at least tion of the American Psychological Association, Los Angeles.
three items, the mean internal reliabilities of these self-reported behav- Davis, M. H., Hull, J. G., Young, R. D., & Warren, G. G. (1987). Emo-
iors in the shorter relationships (mean a - .69 for men and .69 for tional reactions to dramatic film stimuli: The influence of cognitive
women) is virtually the same as for the longer relationships (mean a = and emotional empathy. Journal of Personality and Social Psychol-
.70 for men and .70 for women). In examining the two-item clusters, ogy, 52. 126-133.
the same pattern is found. The mean interitem correlation for these Dean, A., & Lin, N. (1977). The stress-buffering role of social support.
clusters in the shorter relationships (.28 for men, .32 for women) is very Journal of Nervous and Mental Disease, 165, 403-417.
close to that of the longer relationships (.29 for men, .36 for women). Deutsch, F., & Madle, R. (1975). Empathy: Historic and current con-
 410 MARK H. DAVIS AND H. ALAN OATHOUT

capitalizations, measurement, and a cognitive theoretical perspec- profile of marital distress. Journal of Consulting and Clinical Psychol-
tive. Human Development, IS, 267-287. ogy, 4S, 696-703.
Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95, Locke, H. J. (1951). Predicting adjustment in marriage: A comparison
542-575. of a divorced and a happily married group. New York: Holt.
Fiske, S. X, & Taylor, S. E. (1984). Social cognition, Reading, MA: Ad- Neale, M. A., & Bazerman, M. H. (1983). The role of perspective-taking
dison-Wesley. ability in negotiating under different forms of arbitration. Industrial
Franzoi, S. L., & Davis, M. H. (1985). Adolescent self-disclosure and and Labor Relations Review, 36, 378-388.
loneliness: Private self-consciousness and parental influences. Jour- Pedhazur, E. L. (1982). Multiple regression in behavioral research (2nd
nal of Personality and Social Psychology, 8, 768-780. ed.). New York: Holt, Rinehart and Winston.
Franzoi, S. L., Davis, M. H., & Young, R. D. (1985). The effects of
Rusbult, C. E. (1983). A longitudinal test of the investment model: The
private self-consciousness and perspective taking on satisfaction in
development (and deterioration) of satisfaction and commitment in
close relationships. Journal of Personality and Social Psychology, 48,
heterosexual involvements. Journal of Personality and Social Psy-
1584-1594.
chology, 45, 101-117.
Hansson, R. O. (1985, August). Relational competence and adjustment
Solano, C. H., Batten, P. G., & Parish, E. A. (1982). Loneliness and
among the elderly, child abusers, and cancer patients. Paper presented
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

at the annual convention of the American Psychological Association, patterns of self-disclosure. Journal of Personality andSocial Psychol-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Los Angeles. ogy. 43, 524-531.

Hansson, R. Q, Jones, W. H., & Carpenter, B. N. (1984). Relational Wills, T. A., Weiss, R. L., & Patterson, G. R. (1974). A behavioral analy-
competence and social support. In P. Shaver (Ed.), Review of person- sis of the determinants of marital satisfaction. Journal of Consulting
ality and social psychology (Vol. 5, pp. 265-284). Beverly Hills, CA: and Clinical Psychology, 42, 802-811.
Sage. Wright, S. (1934). The method of path coefficients. Annals of Mathe-
Jacobson, N. S., Waldron, H., & Moore, D. (1980). Toward a behavioral matical Statistics, 5. 161-215.

Appendix

Positive and Negative Behavior Items

Positive Behaviors Negative Behaviors

Opening up to the other person (includes readily sharing thoughts, Not opening up to the other person (includes hiding feelings or being
feelings, etc. with the other; disclosing about personal issues; being emo- unemotional or cool, refusing to talk about things the other person may
tionally open, etc.). wish to discuss, etc.).
Readily listening to the other person (includes being willing to listen Not listening to the other person (includes any instance of not being
to the other's thoughts, feelings, etc.; encouraging the other to talk about willing to listen to the other person; would include the other's positive
such matters, etc.). or negative experiences).
Acting in an affectionate or loving way (includes such behaviors as Acting in a rude or insensitive way (includes any inconsiderate behav-
being warm and romantic, gentle and caring, tender and sentimental, ior that does not show courtesy or appreciation for the other person;
etc.). being late, thoughtless, impolite, etc.).
Acting in an appreciative or supportive way (includes reassuring or Acting in a critical or nagging way (includes any nonjoking criticism,
comforting the other person, acting in a way that shows appreciation of mean behavior, teasing, etc.).
the other as a person, etc.). Acting in a dominating or overbearing way (includes giving orders,
Acting in a thoughtful or considerate way (includes any behavior indi- acting bossy, generally trying to dominate, etc.).
cating sensitivity to the wishes or needs of others; being polite, well- Acting in a selfish or egocentric way (includes any tendency not to see
mannered, etc.). anyone else's point of view, being stubborn, boasting, refusing to admit
Acting in a generous or giving way (includes behaviors indicating mistakes, etc.).
helpfulness and unselfishness, willingness to share with others, etc.). Acting in an immature or irresponsible way (includes childishness,
Acting in an even-tempered way (includes any behavior indicating impatience, unreliability, inconsistency, forgetfulness. etc.).
that one is quiet, laid back, not easily angered or upset, etc.). Engaging in an unpleasant personal habit or habits (includes a variety
Acting in a patient or understanding way (includes being tolerant of of crude or repulsive behaviors, such as drinking too much, poor man-
the other, willing to forgive the other, etc.). ners, poor hygiene, foul language, etc.).
Acting in a confident or independent way (includes behaviors indicat- Paying excessive attention to members of the opposite sex (includes
ing self-assurance in life). flirting, having contact with ex-girlfriends or ex-boyfriends, etc.).
Acting friendly or outgoing (includes behavior indicating extraversion Acting in an untrustworthy manner (includes instances of lying,
or having good social skills, enjoying the presence of other people, etc.). cheating on the other person, etc.).
Acting in a positive and optimistic way (includes behaviors indicating Acting in a possessive way (includes being too serious about the rela-
a positive outlook on life, being good-humored and cheerful, being en- tionship, making the other feel smothered, suffocated, etc.).
thusiastic about life, etc.).
Acting in an ambitious or motivated way (includes being dedicated,
determined, a hard worker, etc.).
Acting in an honest or dependable way (includes instances of being
trustworthy, responsible, etc.).
Acting in a wise or intelligent way (includes showing a good common
sense, giving good advice, etc.).

Received October 24, 1986

Revision received February 6, 1987
Accepted March 3, 1987