Journal of Personality and Social Psychology Copyright 1990 by the American Psychological Association, Inc.

1990, \M. 59, No. 6,1238-1249 0022-3514/90/$00.75

The Affective Consequences of Social Comparison:

Either Direction Has Its Ups and Downs
Bram R Buunk Rebecca L. Collins
University of Nijmegen University of British Columbia
Nijmegen, Gelderland, The Netherlands Vancouver, British Columbia, Canada

Shelley E. Taylor Nico W VanYperen

University of California, Los Angeles University of Nijmegen
Nijmegen, Gelderland, The Netherlands
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.

Gayle A. Dakof
University of California, San Francisco

Research on social comparison processes has assumed that a comparison in a given direction
(upward or downward) will lead to a particular affective reaction. In contrast, the present two
studies proposed and found that a comparison can produce either positive or negative feelings
about oneself, independent of its direction. Several factors moderated the tendency to derive posi-
tive or negative affect from upward and downward comparisons. In Study 1, cancer patients low in
self-esteem and with low perceived control over their symptoms and illness were more likely to see
downward comparisons as having negative implications for themselves. Those low in self-esteem
were also more likely to perceive upward comparisons as negative. In Study 2, individuals with high
marital dissatisfaction and those who felt uncertain about their marital relationship were more
likely to experience negative affect from upward and downward comparisons. The implications of
these findings for social comparison theory and for the coping and adaptation literature are dis-
cussed.

In the seminal work on social comparison, Festinger (1954) explanations for these findings, but, in line with Festinger's
suggested that when individuals are uncertain about their opin- theorizing, clearly favored the idea of self-evaluation. More re-
ions or abilities, they will compare themselves with others to cently, social comparison theory has been expanded to include
evaluate their own situation. Schachter (1959) expanded the do- motives for social comparison other than self-evaluation, in-
main of social comparison activities to include emotions. In a cluding self-enhancement (eg., restoring one's self-esteem by
number of experiments, he showed that fear evoked in most comparing oneself with others worse off; Wills, 1981), and self-
subjects the desire to wait with someone else, preferably an improvement (eg., seeking a positive example of the domain
individual in the same situation who reacted with a similar under evaluation; cf. Wilson & Benner, 1971).
degree of emotional intensity. Schachter reviewed a number of The direction of comparison, namely whether one compares
to a better-off or worse-off other (termed upward and down-
ward comparisons, respectively), has been a central part of the
This research was supported by a grant from The Netherlands Orga- theory (Latane, 1966). A great deal of research has substan-
nization for Scientific Research (NWO) to Bram P. Buunk, and by tiated that under conditions in which self-evaluation and self-
National Cancer Institute Grant CA 36409, National Institute of Men- improvement predominate, individuals prefer to compare then-
tal Health Grant MH 42258, and Research Scientist Development state with that of a slightly better-off other (eg, Gruder, 1971;
Award MH 00311 to Shelley E. Taylor. Rebecca L. Collins was sup- Wheeler, 1966; Wilson & Benner, 1971; see also Wheeler et al.,
ported by National Institutes of Health Training Grant MH 15750, and 1969). On the other hand, a substantial body of literature indi-
Gayle A. Dakof was supported by a fellowship from the American
cates that when a comparison is motivated by self-enhance-
Cancer Society—California Division. Nico W VanYperen was sup-
ment, as is the case when self-esteem is threatened, the pre-
ported by a grant from NWO.
We are grateful to Roberta Falke, Jamie Lee, Laurie Skokan, Ina
ferred target of comparison is one who is worse off (Crocker,
Smith, and Robyn Steer for their help in data collection; to Patricia Thompson, McGraw, & Ingerman, 1987; Friend & Gilbert,
Ganz for her help in coding prognosis; and to Monique Cloud for 1973; Hakmiller, 1966; Smith & Insko, 1987; Wills, 1981,1987;
coding the comparison examples. We are also grateful to Damn Leh- Wood, Taylor, & Lichtman, 1985).
man and Joanne Wood for their helpful comments on an earlier ver- In this last line of research, differences in comparison target
sion of this article. selection have been assumed to derive from differences in the
Correspondence concerning this article should be addressed to effects of each type of information. In his downward compari-
Shelley E. Taylor, Department of Psychology, University of California, son theory, Wills (1981) maintained that, under conditions of
405 Hilgard Avenue, Los Angeles, California 90024-1563. threat, downward comparisons are more likely to occur be-
1238
 COMPARISON A N D AFFECT 1239

cause they generate the positive affect essential for self-enhance- Various factors may moderate the affective impact of upward
ment. Downward comparisons appear to boost self-esteem and and downward comparisons. Tesser (1986) proposed relevance
positive emotion and reduce anxiety (Amoroso & Walters, 1969; of the evaluation dimension as a potential moderator. He hy-
Crocker & Gallo, 1985; Gibbons, 1986; Hakmiller, 1966; pothesized that when one is competing with the comparison
Kiesler, 1966; Lemyre & Smith, 1985; Morse & Gergen, 1970). target, comparisons will have the effects typically described in
Upward comparisons appear to be a useful source of self- downward comparison theory, but that in noncompetitive cir-
evaluative information (Nosanchuk & Erickson, 1985; Wheeler cumstances (when the evaluation is not self-relevant), down-
et al., 1969), but seem concurrently to produce negative affect ward comparisons will be negative and upward comparisons
and lower self-evaluations by reminding one that one is inferior will be positive.1 The studies presented here examine addi-
(Diener, 1984; Marsh & Parker, 1984; Morse & Gergen, 1970; tional factors that may produce such outcomes. Study 1 exam-
Salovey & Rodin, 1984; Tesser, Millar, & Moore, 1988; Testa & ined the influence of self-esteem, the likelihood of improve-
Major, 1988). ment or decline on the attribute under evaluation, and per-
In contrast to the previous literature, the present article pro- ceived controllability of the attribute being evaluated. Study 2
poses that the affective consequences of a comparison are not extended these findings by replicating them in a second popula-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

intrinsic to its direction. Although an upward comparison may tion and examining the moderating roles of uncertainty and
This document is copyrighted by the American Psychological Association or one of its allied publishers.

serve the purpose of evaluation more readily than a downward dissatisfaction over the dimension under evaluation.
one, and a downward comparison may more readily serve the First, the effect of comparing may be dependent on personal-
function of self-enhancement, each may not necessarily have ity characteristics of the individual who is making the compari-
this effect. Learning that another is better off than yourself son. Crocker and her colleagues (Crocker & Schwartz, 1985;
provides at least two pieces of information: (a) that you are not Crocker et al., 1987) found that individuals high in self-esteem
as well off as everyone and (b) that it is possible for you to be are more likely to make self-enhancing downward comparisons
better than you are at present. Those able, by virtue of then- than are those withrelativelylow self-esteem. They argued that
personalities or circumstances, to focus on the positive aspect high self-esteem individuals have positive self-concepts, in part,
of this information may feel better about themselves as a result because they engage in these self-enhancing strategies. Their
of an upward comparison. Those who focus on the negative hypothesis assumes that the meaning derived from a compari-
aspect may feel worse. Conversely, learning that another is son is intrinsic to its direction and that downward comparisons
worse off than yourself also provides at least two pieces of always lead to greater self-esteem and upward comparisons do
information: (a) that you are not as badly off as everyone and (b) not. However, a more general version of the hypothesis would
that it is possible for you to get worse. Focusing on the fact that be that individuals high in self-esteem make comparisons favor-
one is better off than others may lead one to feel better about able to themselves, regardless of their objective standing rela-
oneself as a result of a downward comparison, but focusing on tive to the target. Thus, high self-esteem individuals may be
the possibility of getting worse may produce negative feelings more likely to make self-enhancing downward comparisons
about oneself. Thus, how one feels in response to the informa- than low self-esteem persons and more likely to interpret up-
tion that another person is better off or worse off than oneself ward comparisons as self-enhancing as well (cf. Wilson & Ben-
may depend on how one interprets the information. ner, 1971). Conversely, those with low self-esteem may be less
Preliminary evidence suggests that downward comparisons likely to interpret either an upward or a downward comparison
can indeed result in negative feelings. In their studies of victims as favorable to themselves.
of chronic illness (a group under threat to self-esteem), both A second factor that may determine the effect of comparison
Dakof (1986) and Wood et al. (1985) found that these individ- information is the individual's likelihood of improving or de-
uals sometimes felt threatened by exposure to others who had clining on the attribute under evaluation. The importance of
the same disease as themselves, but who were more ill. For the possibility of the comparer attaining the target's level of
example, their respondents described the doctor's waiting achievement for comparison processes has been examined
room as a particularly difficult situation because it forced on previously. Studies testing the related-attributes hypothesis sug-
them the realization that things could be worse. Downward gest that comparisons are more meaningful when the comparer
comparison theory would predict the opposite: These people is similar to the target on dimensions related to that under
should feel better about their own state when they see how evaluation (Goethals & Darley, 1977; Wheeler & Zuckerman,
much better off they are than others (Wills, 1981). Work by 1977). Furthermore, Wheeler (1966) proposed that the com-
Tesser and his colleagues (see Tesser, 1986, for a review) also parer's motivation level affects presumed similarity to the tar-
suggests that downward comparisons can be aversive. When get as well, and thus will determine the choice of a comparison
people learn of worse-off others with whom they are "close" other. Both lines of research imply that the meaning derived
(highly similar or emotionally tied) and the comparison dimen- from a comparison is dependent on the likelihood of finding
sion is not central to self-definition, they may experience nega- oneself at the target's level (Brickman & Bulman, 1977). Thus,
tive affect and arousal. Tesser (1986) and Wood (1984) also if the comparison dimension is a skill acquired through prac-
found evidence for positive affective consequences of upward tice or one that increases naturally with maturity, upward com-
comparisons: People who learned that another had done better
than they had felt better about themselves as a result of this
information. For example, a cancer patient may feel comforted 1
It should be noted that although he uses the same method to assess
or inspired by exposure to another who hasrecoveredfrom the these two processes, Tesser (1986) refers to the latter as a reflection
illness (Taylor & Lobel, 1989). (rather than comparison) process, and distinguishes between the two.
 1240 BUUNK, COLLINS, TAYLOR, VANYPEREN, AND DAKOF

parisons may be uplifting because they provide the comparer of social support needs among cancer patients (Taylor, Falke, Shoptaw,
with the information that such achievements are within reach. & Lichtman, 1986).2 To be eligible for the present study, patients had to
Similarly, an individual who is undergoing a stressful event but be within S years of diagnosis or recurrence and between 30 and 70
whose situation may improve may make an upward compari- years of age. Blocks of potential subjects by gender, estimated progno-
son and feel good, seeing him- or herself as progressing toward sis (good versus fair/poor), and support group membership (yes or no)
the target's superior state. Conversely, for someone whose situa- were constructed for selection purposes.3 Subjects were then randomly
selected from these blocks and invited to participate in the interview
tion is likely to decline, an encounter with another who is worse
study. Of the subjects contacted, 93% agreed to participate (i.e, 4 sub-
off may be threatening. The information may be interpreted as jects declined).
indicative of a worse future, rather than as reassurance about The sample included 30 women and 25 men, ranging in age from 30
one's presently superior state. to 66 with a median age of 54. Eighty-three percent were married, and
Third, the degree to which individuals perceive their prog- 84% had children. Fifty-six percent were employed, and the median
ress as controllable by themselves or by others may affect com- yearly family income was between $40,000 and $49,000. Ninety-three
parison responses. Perceived control has been shown to have percent had completed high school, and 29% were college graduates.
powerful effects on other aspects of cognition (see Fiske & Tay- The sample was 44% Protestant, 25% Jewish, 13% Catholic, and 18%
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

lor, 1984), coping (Taylor, Lichtman, & Wood, 1984; Thomp- had another or no religious affiliation.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

son, 1981), and emotional responses (Thompson, 1981). In Participants had been diagnosed or had sustained a recurrence an
average of 3.2 years prior to the interview (SD =1.7). Twenty percent of
terms of the present theory, people in control may feel that they
respondents were receiving treatment for their cancer at the time of the
have the means to attain a higher level of functioning or avoid a interview. Using medical chart materials, an oncologist rated progno-
downfall, and thus neither downward nor upward comparisons sis on a 5-point scale ranging from very guardedor grave prognosis(1) to
would theoretically pose a threat. In fact, upward comparisons probable cure (5). Thirty-six patients had cancers that were rated 4 or 5
should be inspiring to these individuals. Consistent with this (in remission), and the remainder (19) had prognostic ratings of 1,2, or
point, Testa and Major (1988) found that individuals making 3 (active cancers). Patients with all sites of cancer participated.
upward comparisons reported lower levels of depression and Interview. Respondents were telephoned and the interview was
hostility when control was high than did a group with low con- arranged, usually in the home. At the beginning of the structured
trol. However, they did not have a baseline measure of mood, so interview, respondents received an informed consent form, and per-
it is not clear whether control produced a negative impact in one mission to tape record the interview was obtained. The average inter-
group, a positive impact in the other, or both. view lasted between IVi and 2 hr.
The interview covered basic demographic data, the respondent's
To summarize, the first study was conducted (a) to demon- past and current health status, social support experiences following
strate that comparisons in a given direction can lead to diver- the cancer diagnosis, perceptions concerning how his or her life had
gent affective responses and (b) to determine how the factors of changed following diagnosis, and items relevant to the present investi-
self-esteem, probability of improvement or decline in out- gation, including beliefs about control as well as social comparison
comes, and control over outcomes influence these responses. processes.
The population chosen to examine these issues was cancer pa- Socialcomparisonitems. Fourquestionsconcerningsocialcompari-
tients. Previous research (Wood et al, 1985) has found that the sons related to the present study. Because previous research has already
majority of these individuals make social comparisons, sug- documented the prevalence of comparisons in a similar population
(Wood et al., 1985), we felt closed-ended questions were appropriate.
gesting that it is a prevalent coping strategy among them. This
To avoid leading subjects to report particular affective consequences,
research also found that most comparisons were self-enhancing however, we presented the possibility of both emotional responses
downward comparisons, as would be predicted by downward before questioning subjects more specifically. Downward comparisons
comparison theory (Wills, 1981). Consequently, we expected were assessed first. Subjects were told:
that the majority of comparisons would be self-enhancing, and
that most of these self-enhancing comparisons would be made Some people have told us that when they see cancer patients who
to worse-off others. However, we predicted that when upward are not doing as well as they are, it makes them feel lucky and
comparisons were made, individuals high in self-esteem, those grateful that they are not in worse shape themselves. Other people
have told us that when they see cancer patients who are not doing
who expected their condition to improve, and those who felt as well as they are, it makes them feel worse. For these people,
their future was controllable would be more likely to experi- seeing cancer patients who are worse off only increases their fears
ence them as self-enhancing and less likely to experience them and anxieties.
as aversive than would those low in self-esteem, those who ex-
pected their condition to decline, or those who perceived their Subjects were then asked to rate the frequency with which they had felt
condition as uncontrollable. Similarly, we predicted that per- lucky or grateful when exposed to worse-off others on a 4-point scale
sons with high self-esteem, expectations for improvement, and where 1 = never, 2 = rarely, 3 = sometimes, and 4 = often. Following this,
belief in control would be less likely to experience downward participants giving a rating greater than 1 were asked to provide an
comparisons negatively and more likely to experience them example of a time when they experienced such a comparison. Next,
positively than would those low in self-esteem, with poor prog- subjects were asked how often they had felt fearful or anxious in re-
sponse to such people. Subjects indicated their answer on the same
noses, and with little sense of control.

Study 1 2
In that previous study, the response rate for participation approxi-
Method mated 80%.
3
Support group membership was a blocking variable because it re-
Subjects. The sample consisted of 55 individuals recruited from a lated to other aspects of the interview that dealt with issues of social
pool of 668 cancer patients who had previously participated in a survey support.
 COMPARISON AND AFFECT 1241

4-point scale, and were again asked for a specific comparison instance. comparisons and were associated with the designated affect.
Upward comparisons were assessed next. Subjects were told: Interrater agreement on these judgments was 88%. Inconsisten-
cies were resolved through discussion, and one rater coded the
Some people have told us that when they see cancer patients who
are doing better than they are, it makes them feel frustrated or
remaining cases. Results further substantiate the frequency rat-
depressed. Other people have told us they feel inspired or com- ing data. Eighty-three percent of those who reported experienc-
forted when they see other cancer patients who seem to be doing ing an upward comparison and feeling good (i.e., who re-
better than they are. sponded with a frequency rating greater than never) gave a clear
example of a time when this had happened to them; 71% did so
Again, subjects rated the frequency of each of these reactions on the for downward/positive affect comparisons. Eighty-two percent
same 4-point scale and were asked to provide examples of them.
of those who had experienced downward negative affect com-
Control items. Both personal control and control by external factors
were assessed. The items assessing personal control were prefaced by
parisons also provided a clear example of this. Results were less
"Some people who suffer from serious illness feel that they have some clear for upward comparisons resulting in negative affect: 42%
control over day-to-day symptoms, over the future course of the dis- of those who reported experiencing this were able to provide a
ease, or over the treatments themselves, whereas others do not." Sub- clear instance of its occurrence.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

jects were then asked two questions regarding personal control: (a)
This document is copyrighted by the American Psychological Association or one of its allied publishers.

We also examined the content of these examples. Previous

"Now that your cancer has been detected, to what extent do you feel research (Wood et al., 1985) has identified four major dimen-
you have control over the amount of fatigue, pain, or other symptoms
sions on which cancer patients make social comparisons: prog-
you experience on a daily basis?" and (b) "To what extent do you feel
you can keep the cancer from spreading or coming back?" Subjects
nosis, physical limitations or symptoms, coping and adjust-
indicated their answers on a S-point scale ranging from not at all to ment, and external resources such as finances or social support.
completely. Comparisons in a particular direction and with a particular
Items assessing the control exerted by other factors were prefaced consequence may be more likely to be made on some dimen-
with "Some people believe that someone or something else may have sions than others (e.g., it may be easier to feel good about some-
control over day-to-day symptoms, over the future course of disease, or one with better adjustment than someone with superior re-
over treatments themselves, whereas others do not." This was followed sources to one's own). An independent rater coded each compar-
by two questions: (a) "To what extent do you think someone or some- ison example for the dimension involved so that this possibility
thing else has control over your day-to-day symptoms?" and (b) "To could be tested. Interrater reliability (again based on a subset of
what extent do you feel that someone or something else can keep your 25% of the cases coded by an additional rater) for this measure
cancer from spreading or coming back?" Each of these items was fol- was 90%. We then conducted four Cochran Q tests, one for each
lowed by the same scale ranging from not at all (1) to completely (5).
comparison/affect item. The Cochran Q tests whether the pro-
Self-esteem. The Rosenberg Self-Esteem Inventory (Rosenberg,
portions in dichotomous categories (in this case, mentions ver-
1965) was used as a measure of chronic self-esteem. This questionnaire
was left with the patient at the conclusion of the interview, along with
sus nonmentions of a comparison dimension) are the same
measures of adjustment and indicators of cancer-related problems that across variables (in this case, across the four dimensions). There
were part of the larger study. Eighty percent of the respondents re- were significant differences in the frequency with which each
turned this questionnaire. dimension was used. This occurred in response to all four ques-
tions (downward/negative affect Q = 27.01, p < .001; down-
ward/positive affect Q -15.40, p = .002; upward/negative affect
Results Q = 9.00, p = .03; upward/positive affect Q = 16.67, p < .001).
Looking at the number of respondents who had ever experi- However, examination of the frequencies reveals that the pat-
enced each type of comparison paired with each affective con- terns of response were the same for three of the four questions.
sequence, 82% (n = 42) of subjects made downward compari- For all but upward/positive affect comparisons, comparisons of
sons and felt good, 59% (n = 28) made downward comparisons prognosis were the most frequent (ranging from 52% to 67% of
and felt bad, 40% (« = 19) made upward comparisons and felt each category); comparisons on other dimensions were infre-
bad, and 78% (n = 37) made upward comparisons and felt quent and occurred about equally often. In the case of upward
good. 4 Although a substantial proportion of the sample re- comparisons resulting in positive affect, comparisons of prog-
ported making each type of comparison with each effect, there nosis were also most common (50%); however, comparisons of
were differences in the frequency with which each comparison adjustment were nearly as frequent (33%). Comparisons of re-
was reported to have occurred. A one-way multivariate analysis sources and physical status were, again, infrequent and approxi-
of variance (ANOVA) with four repeated measures of compari- mately equal.
son frequency produced a significant effect, ,F(3,135) = 13.87, It was hypothesized that high self-esteem individuals would
p < .001. Downward positive affect comparisons were the most make more positive affect and less negative affect comparisons
common, as predicted, although upward positive affect compar- in both directions than would low self-esteem individuals. To
isons were almost equally common. As expected, downward test this, the sample was split into two groups, those above and
negative and upward negative affect comparisons occurred less those below the median score on the self-esteem inventory.
often. Patients reported making comparisons resulting in posi- Four t tests were conducted comparing the frequency with
tive affect, regardless of direction, more frequently than com-
parisons resulting in negative affect, f(52) = 6.09, p < .001.
Mean frequency ratings and simple comparisons are presented 4
Percentages reported are of subjects responding to the question. A
in Table 1.
few respondents indicated that they never compared themselves with
Two independent raters coded 25% of the cases to determine others, and some indicated that they did not know any cancer patients
whether the examples provided by subjects clearly constituted with whom the comparison could be made.
 1242 BUUNK, COLLINS, TAYLOR, VANYPEREN, AND DAKOF

Table 1 much previous theorizing and empirical emphasis, the data

Mean Frequency Ratings of Upward and Downward revealed that upward and downward comparisons are not in-
Comparisons Paired With Positive and Negative Affective trinsically linked to particular affective outcomes. Instead, as
Consequences Among Cancer Patients predicted, we found that people may construe both upward
and downward comparisons as either positive or negative.'
Affective consequence
Consistent with downward comparison theory (Wills, 1981),
Type of comparison Positive Negative the present study found that cancer patients most frequently
engaged in self-enhancing downward comparisons. We had
Upward 2.63. 1.60.,, proposed, however, that upward comparisons could be inter-
Downward 2.75 b 2.00.,, preted in a self-enhancing manner as well. This hypothesis was
Note. Means sharing a common subscript are significantly different,
supported. In fact, a majority of the sample made self-enhanc-
p s .01. Scales ranged from 1 (fiever) to 4 (often). N= 55. ing upward comparisons, and these occurred nearly as often as
self-enhancing downward ones. For example, one of our re-
spondents said:
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.

which each group reported making comparisons in each direc- When I was going through the worst period, you know, the acute
time of therapy and stuff, it was gratifying to see people recover-
tion with each affective consequence (upward/positive affect, ing, having their hair grow back, getting their strength, and so on.
downward/positive affect, upward/negative affect, and down- Yeah, that was very positive and very helpful.
ward/negative affect). There were significant differences in the
frequency with which persons high and low in self-esteem re- Just as both directions of comparison may be interpreted
ported comparisons resulting in negative affect. Specifically, positively, both may also lead to negative self-perceptions. In-
high self-esteem individuals were less likely to feel bad when formation that another person is doing worse than oneself can
comparing upward (M frequencies = 1.25 and 1.95), f(31) = be depressing, as can information that someone is doing better
2.60, p = .007, one-tailed, and less likely to feel bad when com- than oneself. Interestingly, in the present study the former took
paring downward (M frequencies = 1.60 and 2.43), ;(39) = 2.50, place more often than did the latter: Downward comparisons
p = .008, one-tailed, than were low self-esteem individuals. The more frequently led to negative affect than did upward compari-
two groups did not differ in the frequency with which they sons. Respondents tended to be frightened by the experiences
reported positive affect in response to comparison, both ps of patients who were notrecoveringfrom their cancer. For ex-
(one-tailed) > . 10. ample, one respondent, a breast cancer patient who had been
Prognosis was also hypothesized to influence interpretation treated by lumpectomy, said:
of comparison information. Participants were divided into a Well, my girl friend who had the second mastectomy, she has really
good prognosis group (ratings of 4 and 5) and a poor prognosis been through the mill. . .it raises my anxiety terribly because she
group (ratings of 1 through 3). Four t tests were conducted to started off with a lumpectomy and radiation.
test for differences between these groups in the frequency with
which they reported each comparison affect and direction. This ability to derive positive or negative affective consequences
There were no significant differences found, all ps (one- from social comparisons was, as evidenced byrespondents'ex-
tailed) > .05. amples, not specific to comparisons of a particular aspect of
Finally, it was hypothesized that individuals who believe they one's cancer, such as prognosis or adjustment. There was, how-
or others have control over their illness would be more likely to ever, a tendency for upward comparisons on the dimension of
report feeling good in response to upward comparisons and coping to more often lead to positive affect than those made on
less likely to report feeling bad in response to downward com- other dimensions.
parisons, as compared with those with lower perceived control. The second important set of findings for social comparison
Correlations were calculated between each rating of control
(personal and other control over daily symptoms and the future 5
One might conjecture that high self-esteem would be related to
course of the illness) and the frequency of each type of compari-
perceived control and, therefore, that the two sets of analyses would be
son. Because prognosis may be related to perceived control, it redundant. However, this was not the case. Self-esteem was unrelated
was partialed out of these analyses. As predicted, the belief that to either perceived control over daily symptoms (r = . 10) or perceived
one has personal control over the future course of one's illness control over the disease (r = . 13).
and the belief that one has control over daily symptoms were 6
Rather than interpreting the same information differently, an alter-
both inversely associated with feeling bad in response to worse- native process of deriving particular effects from comparisons would
off others (rs = -.31 and -.30, respectively, both ps < .05). entail the selection of either upward or downward comparison targets
Perceived other control was unrelated to this measure, and no who are likely to reflect on oneself in a particular way (e.g., Tesser's,
type of control was associated with the frequency of upward 1986, theory suggests the choice of similar or dissimilar targets). How-
positive affect comparisons.5 ever, it is clear from the examples provided by subjects in the present
study that many of them were interpreting exactly the same compari-
son information in different ways. Many referred to the same target in
Discussion response to questions assessing both positive and negative reactions. In
addition, when characteristics of the target were provided, they were
The present study produced four sets of results that have often the same across positive and negative questions, even when the
relevance to social comparison theory. First, in contrast to specific target was different.
 COMPARISON AND AFFECT 1243

theory concerned the role of moderating factors in the affective (namely uncertainty about the quality of one's marriage and
consequences of social comparisons. The ability to avoid nega- marital dissatisfaction).
tive comparisons was more true of individuals high in self-es- Specifically, by examining uncertainty about how things
teem and (for downward comparisons) those who believed they were going in one's marriage, we were able to assess the impact
could control the symptoms or course of the disease than of of uncertainty on social comparison processes. Festinger (1954)
people low on these characteristics. With respect to self-es- had suggested that the desire for social comparison is particu-
teem, we had hypothesized that high self-esteem individuals larly strong in the case of uncertainty about one's opinions or
would be more likely to make self-enhancing comparisons in abilities. Schachter (1959) maintained that the desire for self-
either direction, and less likely to experience comparisons as evaluation (i£., the desire to evaluate the appropriateness of
threatening, than would persons lower in self-esteem. Only the one's reactions) is the primary motive underlying the tendency
latter half of this hypothesis was supported. Whereas self-es- to affiliate under stress. Taken together and in conjunction
teem did not affect the frequency with which comparisons were with the data from Study 1, these points suggest that uncer-
seen as positive, low self-esteem persons were significantly tainty may predispose people to be especially vigilant to com-
more likely to experience negative outcome comparisons than parison information generally and to negative information
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

were persons of high self-esteem. more specifically. We therefore predicted that those uncertain
This document is copyrighted by the American Psychological Association or one of its allied publishers.

A similar effect was found with regard to psychological con- in their marriages would report making more social compari-
trol. Those people who felt they could control their symptoms sons of all kinds and would be more likely to focus on the
and the future course of the illness were less likely to feel threat- negative than the positive aspects of upward and downward
ened by exposure to very ill patients. They were not, however, comparisons.
any more likely to focus on the positive implications of an- In an effort to create a situational analogue to self-esteem, we
other's successful recovery than were persons who felt their also examined marital satisfaction as a determinant of the af-
future health was uncontrollable.7 fect generated by upward and downward comparisons. We pre-
Third, the ability to avoid negative comparisons appears to dicted that, like the low-esteem individuals reacting to self-re-
depend more on the subjective than objective characteristics of levant comparison information in Study 1, individuals experi-
the threatening event. We had predicted that prognosis would encing marital dissatisfaction might be more vigilant to the
influence whether one derived hope from the knowledge that negative information suggested by comparisons of their own
others were improving and despair at news of another's decline. marriage to that of others. We therefore expected those high in
However, the comparer's prognosis was not related to the fre- marital dissatisfaction to make more negative affect compari-
quency with which a comparison was seen as threatening or sons and fewer positive affect comparisons than those higher in
enhancing. The rating of prognosis was an objective measure, marital satisfaction.
however, and many cancer patients retain the belief that they
will recover in spite of indications to the contrary. Indeed, Method
Wood and her colleagues (Wood et al, 1985) also failed to find
an effect of prognosis when using an objective measure to pre- Subjects. The sample consisted of 632 married individuals, 304
dict enhancing downward comparisons (see also Marks, men and 328 women. Of the sample, 432 subjects (69%) were recruited
Richardson, Graham, & Levine, 1986). Our positive findings using an announcement placed in local newspapers, the remaining 200
concerning control suggest that a subjective measure of respon- (31%) by contacting a random sample of a middle-sized Dutch town.
dents' perceived prognosis might have been associated with the This last sample was provided by the city council. The average length
of marriage was 16.4 years (range: less than 1 month to 55.3 years). The
ability to derive benefit from comparisons.
mean age was 40.6 years (range: 21 to 81), and 79% of the sample had
Finally, self-serving social comparisons appear to mute the children. Highest level of education achieved varied from elementary
effect of negative information rather than enhance available education (3.6%) to college education at the master's level or higher
positive information; that is, both self-esteem and control were (10.9%). Of the sample, 83% of the men and 32% of the women were
associated with fewer negative affect comparisons, but were employed outside the home for 20 or more hours per week. A wide
unrelated to positive comparisons. range of occupations were represented in the sample. The number of
subjects in the following analyses varies because of occasional missing
data.
Procedure. The data on the 434 subjects mentioned above were
Study 2 collected as part of the second measurement of a longitudinal study on
social comparison processes in marriage (VanYperen & Buunk, 1990).
Although Study 1 demonstrated that comparisons in a given The subjects were contacted by mail and were asked to complete, in
direction can produce divergent affective responses, it did so privacy, an anonymous questionnaire about "marital relationships."
within a select population, cancer patients. Recovery is possi- They were not to discuss this questionnaire with their partner before
ble, but most patients are at least somewhat uncertain about completing it. After 2 weeks, nonrespondents received areminderand
their prognosis, and unexpected changes in physical states may
take place. This factor may influence the data patterns ob- 7
tained. Study 2 provided an opportunity to replicate the pat- Somewhat surprisingly, the perception that another agent con-
trolled the course of the illness was unrelated to comparison effects.
terns of comparison responses obtained in Study 1 in a different Previous research (Taylor, Lichtman, & Wbod, 1984) found that both
population (Dutch married people) using a different compari- vicarious and personal control were associated with positive adjust-
son dimension, namely quality of one's marital relationship, ment to cancer. Apparently, vicarious control does not have the same
with different but conceptually related moderating variables impact on self-serving social comparisons.
 1244 BUUNK, COLLINS, TAYLOR, VANYPEREN, AND DAKOF

after 4 weeks, a second reminder as well as a new questionnaire. The Table 2

same procedure was followed for a randomly selected sample of 1,000 Mean Frequency Eatings of Upward and Downward
individuals from a middle-sized Dutch city. Of this sample, a total of Comparisons Paired With Positive and Negative Affective
200 persons (20%) sent back the questionnaire, making up the remain- Consequences Among Married Individuals
ing 31 % of the total sample for this study. To obtain an indication ofthe
reasons for nonparticipation, a brief questionnaire was sent to the Affective consequence
nonsample. This questionnaire was sent back by 32% of these individ-
uals. The reasons mentioned most often for nonparticipation were Type of comparison Positive Negative
"My private life is my own business" (17%), followed by "I get annoyed
when I receive a questionnaire without having asked for it" (13%), Upward 2.70 1.87
"I just don't feel like it" (6%), and "I am too old for these kinds of Downward 3.59 1.62
things" (5%).
Note. All differences among means are significant, p .001. Scales
Social comparison items. To assess the affect evoked by upward and ranged from 1 (never) to 5 fyery often). N= 632.
downward comparisons, the subjects were presented with modified
versions of the four questions from Study 1. To measure the frequency
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

of positive affect evoked by downward comparisons, subjects were

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

asked to answer the following question (translated from Dutch): "How

often do you feel happy and pleased when you compare your own ence was that negative affect upward comparisons were re-
marital relationship with that of others who have a relationship that is ported more often by women (M = 1.95) than by men (M =
worse than yours? A 5-point scale was used, with points labeled 1 1.79), /(614) = 2.30, p < .05. That is, women more often than
(never), 2 (seldom), 3 (sometimes), 4 (quite often), and 5fyeryoften). Tomen felt bad when they learned about happier couples.
assess the frequency of negative affect evoked by downward compari- Marital dissatisfaction and comparisons. We had predicted
sons, subjects were asked: "How often do you feel unhappy and dis- that individuals with high marital dissatisfaction would make
pleased when you compare your own marital relationship with that of fewer positive and more negative comparisons in both direc-
others who have a relationship that is worse than yours?" Similar ques- tions, whereas the opposite pattern would be characteristic of
tions were asked for negative and positive affect evoked by marriages individuals with satisfying marriages. In order to test this pre-
that were better than one's own.
diction, respondents were divided into three groups on the ba-
Uncertainty item. To assess the degree of uncertainty in marriage, sis of their marital dissatisfaction scores. Because of ties in the
the respondents were asked to indicate on a 5-point scale (1 = not at all,
scores on this scale, it was not possible to divide them into
2 = hardly, 3 = somewhat, 4 = quite, and 5 = very much) to what extent
they felt uncertain about how things were going in their marriage. groups of equal size (for the high dissatisfaction group [scores
Marital dissatisfaction. An 8-itera scale (Buunk, 1990) was used to 1.39 or lower], n - 145; for the medium dissatisfaction group
assess marital dissatisfaction. This scale has proven to have high reli- [scores between 1.39 and 1.90], n = 254; and for the low dissatis-
ability and stability and to relate meaningfully to other variables, in- faction group [scores 1.90 or higher], n = 198).
cluding the stability of the relationship and self-rated coping with mar- Examining the effects of marital dissatisfaction on upward
ital problems (e.g., Buunk, 1987; Buunk & Bosnian, 1986). Respondents and downward comparisons, a MANOVA with three levels of
indicated how often an item applied to their marital relationship on marital dissatisfaction was conducted, using the four social
the same 5-point scale that was used for the social comparison items, comparison items as dependent measures. Analyses were con-
ranging from never (1) to very often (5). Five of the items referred to
ducted separately for each gender, inasmuch as previous litera-
negative feelings and behaviors, such as "My partner irritates me" or
"We have quarrels." Three items referred to positive experiences such ture suggests that men and women respond differently to
as "Things are going well between us." Coefficient alpha of the scale sources of strain inrelationships(Titus, 1980). The multivariate
was .85 in this sample. Marital dissatisfaction and uncertainty were effect ofrelationshipdissatisfaction on social comparisons was
correlated .50 in the sample. significant for men, F(8, 556) = 7.33, p < .001, as well as for
women, F(8,582) = 9.94, p < .001.
Univariate analyses showed that the level of marital dissatis-
Results
faction was significantly and strongly related to the frequency
Frequency ofcomparisons. Table 2 makes clear that, as was with which downward negative affect comparisons were re-
the case in Study 1, positive affect comparisons were made ported: for men, F(2, 280) = 10.96, p < .001; for women, F(2,
more frequently than comparisons evoking negative affect. 293) = 15.62, p < .001. The frequency of upward negative effect
Downward comparisons generating positive affect were the comparisons was even more strongly related to level of marital
most common. Nearly all subjects (95%) reported such compari- satisfaction: for men, F{2, 280) = 26.62, p < .001; for women,
sons, and 59% said they made this type of comparison quite or F(2, 293) = 33.23, p < .001. As Figure 1 shows, the higher the
very often. Upward positive affect comparisons also occurred level of marital dissatisfaction, the more often individuals felt
regularly among the subjects: 78% of subjects reported experi- unhappy and bad when they compared themselves with cou-
encing these at some time; 28% stated they made such compari- ples who had better marriages, and the more often marriages
sons quite or very often. Negative affect upward or downward worse than their own also evoked negative affect.
comparisons were less frequent; 59% of the sample stated they In contrast, marital dissatisfaction had little influence on the
made upward comparisons that made them unhappy some- frequency with which comparisons to other couples evoked pos-
times, and only 3% made such comparisons quite or very often. itive affect (see Figure 1). There was no significant difference in
About half of the sample (48%) reported downward compari- the frequency with which positive affect upward comparisons
sons evoking negative affect, and just 2% reported they made were made. However, among women and marginally among
such comparisons quite or very often. The only gender differ- men, there was an effect of marital dissatisfaction on the fre-
 COMPARISON AND AFFECT 1245

Upward Comparison Downward Comparison

4-

Q>
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.

low moderate high low moderate high

level of marital dissatisfaction level of marital dissatisfaction

males positive affect

males negative affect
females positive affect
females negative affect

Figure 1. Positive and negative affect generated by upward and downward comparisons
for various levels of marital dissatisfaction.

quency of positive affect downward comparisons: for women, The univariate F tests confirm this suggestion. Uncertainty
F(2, 293) = 3.99, p < .05; for men, F(2, 280) = 2.64, p < .07. was clearly related to the frequency of negative affect downward
People in less happy marriages felt less positively in response to comparisons: among men, F(2,291) = 11.33, p < .001; among
downward comparisons than those whose marriages were women, F(2, 331) = 11.71, p < .001. As Figure 2 shows, the
more satisfying. more uncertain individuals felt about their own marriages, the
Marital uncertainty and comparisons. On the basis of re- more they experienced negative affect when confronted with
sponses to the uncertainty item, three groups were created, the marriages that were worse. Comparisons with better marriages
first consisting of those who said they were not at all uncertain appear to generate negative affect even more often. The rela-
about how their marriage was going (score 1; n = 199), the sec- tionship between uncertainty and the frequency of negative af-
ond consisting of individuals who said they were hardly uncer- fect upward comparisons was strong and significant among
tain (scores 2; n = 275), and the third consisting of those who men, F(2,291)=38.09, p<. 001, as well as among women, F(2,
indicated that they felt somewhat, quite, or very uncertain 331) = 40.91, p < .001. As is apparent from Figure 2, people
(scores 3,4, and 5; n = 157). who felt uncertain about how things were going in their
We had predicted that uncertainty would increase compari- marriages were unhappy and dissatisfied more often when
sons of all types and that those high in marital uncertainty they compared their own marriage with happier marriages. No
would make fewer positive and more negative comparisons significant effects of uncertainty were found with respect to the
than those low in uncertainty. Overall, those high in uncer- positive affect comparisons.
tainty made more social comparisons than those low in uncer-
tainty. A MANOVA using three levels of uncertainty as the
independent variable and the social comparison items as de- General Discussion
pendent variables yielded significant main effects for uncer- Taken together, the findings from two studies examining
tainty, F(8,578) = 9.30, p < .001, for men and F(8,618) = 9.55, social comparison processes in different domains have impor-
p < .001, for women. However, examination of Figure 2 sug- tant implications for social comparison theory. Both studies
gests that this effect can be accounted for primarily by negative demonstrated that comparison direction (upward or down-
affect comparisons. ward) is not intrinsically linked to affect, as the previous litera-
 1246 BUUNK, COLLINS, TAYLOR, VANYPEREN, AND DAKOF

Upward Comparison Downward Comparison

4T 1 4'
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.

low moderate high low moderate high

level of uncertainty
level of uncertainty

males positive affect

males negative affect
females positive affect
females negative affect

Figure 2. Positive and negative affect generated by upward and downward

comparisons for various levels of uncertainty.

ture has often assumed. Rather, both upward and downward comparison information differently from people in normal situ-
social comparisons are capable of generating positive or nega- ations.
tive affective responses, depending on which aspect of the com- Another significant regularity in the data concerns the fact
parison is focused on. that the moderating variables of self-esteem and controllability
The results from the two studies also show some commonali- (in Study 1) and uncertainty and marital satisfaction (in Study 2)
ties in the frequencies of different kinds of comparisons. Both largely affected the frequency of negative affect but not positive
studies found that positive affect comparisons were the most affect comparisons. Taken together, these results tie in with a
common comparisons reported, with positive downward com- larger body of literature in social cognition suggesting that
parisons more common than positive upward comparisons. there are cognitive filters of selective attention, representation,
There are some differences in frequency of types of compari- and recall that help people maintain positive beliefs (see Taylor
sons between the two studies. Whereas in Study 1, upward com- & Brown, 1988, for a review). Consistent with that body of data,
parisons evoked positive affect nearly as often as downward the results suggest that these filters operate more to keep the
comparisons, in Study 2 upward comparisons were less potent negative implications of information out of view than to en-
in generating this effect than were downward comparisons. In hance available positive information.
the same vein, in Study 2, upward comparisons leading to nega- The results concerning self-esteem provide an interesting in-
tive affect were more frequent than downward comparisons sight into a current issue in the comparison literature. Our find-
leading to negative affect, whereas in Study 1 the reverse was ings are somewhat inconsistent with Wills's (1981) downward
true. Perhaps this difference depends on the two types of comparison theory, which predicts that low self-esteem individ-
events, namely cancer and marriage. In normal situations (such uals or individuals under threat should make a greater number
as the evaluation of an ongoing marriage), upward comparisons of self-enhancing downward comparisons. Overall, individuals
may be relatively threatening because they remind people how who evaluated themselves (Study 1) or their situation (Study 2)
poorly they are doing, whereas in stressful circumstances (such negatively made substantial numbers of downward compari-
as undergoing cancer), seeing a positive example may be encour- sons, but as just noted, they did not derive a greater amount of
aging and inspiring. Further research is needed to determine self-enhancing information as a result. This finding is more
whether individuals experiencing an unusual threat interpret consistent with Crocker et al.'s (1987) results, which found that
 COMPARISON AND AFFECT 1247

high self-esteem individuals were better able than low self-es- automatic response bias was involved. Third, reports of affec-
teem individuals to make comparisons that are self-serving. tive consequences of particular comparisons varied systemati-
Crocker et al.'s data had demonstrated that high self-esteem cally with perceived control in Study 1 and with dissatisfaction
people make downward comparisons for this purpose. The pres- and uncertainty in Study 2, which would not be expected from
ent results suggest that high self-esteem individuals are better a simpleresponsebias. Moreover, the question of how to inter-
able to make use of either upward or downward comparisons pret positivity in psychologicalresponsesis itself under debate.
for the purpose of self-enhancement than are low self-esteem Rather than representing a response bias, many psychologists
individuals. have argued that mild positivity is how the majority of people
We had predicted that uncertainty would increase compari- experience a broad array of outcomes and that responses indi-
sons of all kinds, a prediction that was not upheld. Instead, in cating such are not themselves a function of response set, but
Study 2, uncertainty and dissatisfaction regarding one's accuratelyreflecta mildly positive perception of the world (eg.,
marriage related to affective consequences of social compari- see Parducci, 1968; Taylor & Brown, 1988).
sons the same way, namely in terms of more frequent negatively The present data have implications for certain long-standing
valenced comparisons. There are at least two possible explana- issues in the social comparison and coping literatures, particu-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tions for thisresult.One is to argue that uncertainty and dissatis- larly how people respond to and cope with forced comparisons
This document is copyrighted by the American Psychological Association or one of its allied publishers.

faction are tapping the same construct. The two measures were (Brickman & Bulman, 1977; Mettee & Smith, 1977). The com-
highly correlated (SO). Arguing against this point is the fact that parison environment appears to be somewhat less malleable
marital dissatisfaction also increased the frequency of positive than was characterized by Festinger (1954; see Wood, 1989). As
affect downward comparisons, a finding that was not mirrored a result, comparison targets are sometimes forced on the com-
in the uncertainty data.8 The other explanation maintains that parer, as in the case of cancer patients exposed to other patients
these measures are tapping two separate dimensions, negativity in the waiting room. The present analysis suggests that some
and ambiguity, that have been previously identified in the stress people, particularly those who evaluate themselves or their situ-
literature as enhancing the perception of stress (e.g., Billings & ation positively or those with a sense of personal control, may
Moos, 1984; Gal & Lazarus, 1975; Holahan & Moos, 1986; respond to unwanted comparisons much as they respond to
McFarlane, Norman, Streiner, Roy, & Scott, 1980; Myers, Lin- other negative information in their environments,filteringand
denthal, & Pepper, 1972; Sarason, Johnson, & Siegel, 1978; Sto- distorting the data to fulfill their needs and expectations (see
kols, Ohlig, & Resnick, 1978; Vinokur & Selzer, 1975). Uncer- Taylor & Brown, 1988, for a review).
tainty and dissatisfaction may both increase the experience of The results also have implications for the literature on coping
strain, which in turn may increase vigilance to the negative and adaptation. Both Wills (1981) and Taylor and Lobel (1989)
information inherent in social comparisons. If uncertainty and have assumed that the propensity to make downward compari-
dissatisfaction are indeed two separate dimensions of marital sons under threat stems from an augmented need for self-en-
perceptions, dissatisfaction would seem to create a greater need hancement induced by threat. Yet the results from Study 2 sug-
for self-enhancing downward comparisons than uncertainty/ gest that negative affect comparisons in both directions are
ambiguity. especially augmented by marital dissatisfaction and that posi-
There are limitations to the studies. The direction of causal- tive affect downward comparisons are slightly lower among
ity cannot be determined for the moderating variables. Those those high in marital dissatisfaction relative to those low in
high in self-esteem, control, marital satisfaction, or certainty marital dissatisfaction. Recall, too, that in Study 1 (in which all
may make different comparisons than others, or, alternatively, subjects were under some degree of threat), negative affect
the avoidance of threatening comparisons may result in a more downward comparisons were relatively more prevalent than in
positive self-image, elevated perceptions of control, greater cer- Study 2. It appears that the effects of threat on social compari-
tainty in one's perceptions, or greater satisfaction. As noted son may be more complex than has been previously assumed.
earlier, past research has demonstrated that the use of self-en- Although threat may produce a propensity for self-enhancing
hancing downward comparisons does improve self-esteem downward comparisons, it may simultaneously increase all
(Crocker & Gallo, 1985; Lemyre & Smith, 1985; Morse & Ger- kinds of negative affect comparisons, an effect that may aug-
gen, 1970). The avoidance of threatening comparisons may
have the same effect: Avoiding negative comparisons may be
one way in which people high in self-esteem, control, satisfac- 8
However, the single item measuring uncertainty may have been a
tion, or certainty maintain these perceptions.9 lessreliableindicator of a common underlying dimension than the
The question arises as to whether theresultsfrom both stud- multiple items assessing marital dissatisfaction. This could explain
ies can be interpreted as evidence of a positivity response bias. why marital dissatisfaction, as the more sensitive indicator, would be
There is considerable evidence in the literature that people are associated with positive affect downward comparisons, whereas un-
biased toward perceiving events positively, a phenomenon that certainty, as a relatively insensitive indicator, might not.
9
Matlin and Strang have termed the Pollyanna principle (Matlin In the case of the control results, it seems unlikely that the avoid-
& Strang, 1978). Several factors argue against a response bias ance of negative information would lead to a stronger belief in per-
sonal control, except through the mediating variable of self-esteem.
interpretation. First, there is little evidence of a positivity bias People who use comparisons to maintain the belief that they are capa-
in these data; rather, negativity is avoided. Second, subjects who ble may infer that they are also in control of their situations. However,
reported particular affective consequences of particular com- self-esteem and control were unrelated in Study 1. It thus seems more
parisons generally had examples readily available to buttress likely that the belief in personal control enables one to avoid the nega-
their perceptions, a finding that suggests that more than an tive impact of comparison information rather than the reverse.
 1248 BUUNK, COLLINS, TAYLOR, VANYPEREN, AND DAKOF

ment rather than diminish distress. Of relevance too is the find- Friend, R. M., & Gilbert, J. (1973). Threat and fear of negative evalua-
ing that persons who feel relatively less control over their health tion as determinants oflocus ofsocial comparison. Journal of Person-
may also be threatened by downward comparison information, ality. 41, 328-340.
rather than comforted by it, as previous theory and results have Gal, R., & Lazarus, R. S. (1975). The role of activity in anticipating and
suggested (Wills, 1981; Wood et al, 1985). Finally, the results of confronting stressful situations. Journal ofHuman Stress, 1, 4-20.
Study 2 suggest that those high in uncertainty, and therefore Gibbons, F. X. (1986). Social comparison and depression: Company's
likely to seek comparison information, are also more likely to effect on misery. Journal of Personality and Social Psychology, 51,
feel threatened by what they learn. The conditions that increase 140-148.
or decrease threatening interpretations of comparisons clearly Goethals, G. R., & Darley, J. M. (1977). Social comparison theory: An
attributional approach. In J. M. Suls & R. L. Miller (Eds.), Social
merit additional study.
comparison processes: Theoretical and empirical perspectives (pp.
Finally, thefindingthat both upward and downward compar- 259-278). Washington, DC: Hemisphere.
isons can be used for the purpose of self-enhancement ad- Gruder, C. L. (1971). Determinants of social comparison choices. Jour-
dresses a long-standing question in the literature on social com- nal ofExperimental Social Psychology, 7, 473-489.
parisons: How can people who are in need of self-enhancing
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Hakmiller, K. L. (1966). Threat as a determinant ofdownward compar-

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

feedback make use of better-off others to facilitate eventual ison. Journal of Experimental Social Psychology, Suppl. I, 32-39.
change in their standing? If people are not capable of so doing, Holahan, C. J., & Moos, R. H. (1986). Personality, coping, and family
the preservation of self-esteem could have negative long-term resources in stress resistance: A longitudinal analysis. Journal of
consequences, leading people to ignore strategies of improve- Personality and Social Psychology, 51, 389-395.
ment. Our results address this concern, suggesting that people Kiesler, S. B. (1966). Stress, affiliation and performance. Journal of
can make use of comparisons in either direction in order to Experimental Research in Personality, 1, 227-235.
simultaneously provide useful information and to maintain Latane, B. (1966). Studies in social comparison: Introduction and over-
their positive self-perceptions (cf. Taylor & Lobel, 1989). view. Journal ofExperimental Social Psychology, Suppl. 1,1-5.
Lemyre, L., & Smith, P. M. (1985). Intergroup discrimination and self-
esteem in the minimal intergroup paradigm. Journal ofPersonality
References and Social Psychology, 49, 660-670.
Marks, G., Richardson, J. K., Graham, J. W, & Levine, A. (1986). Role
Amoroso, D. M, & Walters, R. H. (1969). Effects of anxiety and so- of health locus of control beliefs and expectations of treatment effi-
cially mediated anxiety reduction on paired-associate learning. cacy in adjustment to cancer. Journal ofPersonality and Social Psy-
Journal of Personality and Social Psychology, 11, 388-396. chology, 51, 443-450.
Billings, A. C, & Moos, R. H. (1984). Coping, stress, and social re- Marsh, H. W, & Parker, J. W (1984). Determinants of student self-con-
sources among adults with unipolar depression. Journal ofPersonal- cept: Is it better to be arelativelylarge fish in a small pond even if
ity and Social Psychology, 46, 877-891. you don't learn to swim as well? Journal of Personality and Social
Brickman, P., & Bulman, R. J. (1977). Pleasure and pain in social com- Psychology. 47, 213-231.
parison. In J. M. Suls & R. L. Miller (Eds.), Social comparison pro- Matlin, M. W, AStrang, D. J. (1978). The Pollyanna principle: Selectiv-
cesses:Theoreticalandempiricalperspectives(ppA49-186).'Washm$- ity in language, memory, and thought. Cambridge, MA: Schenkman.
ton, DC: Hemisphere. McFarlane, A. H., Norman, G. R., Streiner, D. L., Roy, R., & Scott, D. J.
Buunk, B. (1987). Conditions that promote break-ups as a consequence (1980). A longitudinal study of the influence of the psychosocial
of extradyadic involvements. Journal of Social and Clinical Psychol- environment on health status: A preliminary report. Journal of
ogy, 5, 237-250. Health and Social Behavior, 21,124-133.
Buunk, B. (1990). Relationship interaction satisfaction scale. In J. Tou- Mettee, D. R., & Smith, G. (1977). Social comparison and interpersonal
liatos, B. F. Perlmutter, & M. A. Straus (Eds.), Handbook of family attraction: The case for dissimilarity. In J. M. Suls & R. L. Miller
measurement techniques (pp. 106-107). Newbury Park, CA: Sage. (Eds.), Social comparison processes: Theoretical and empirical per-
Buunk, B., & Bosman, J. (1986). Attitude similarity and attraction in spectives (pp. 69-101). Washington, DC: Hemisphere.
marital relationships. Journal of Social Psychology, 126,133-134.
Morse, S., & Gergen, K. J. (1970). Social comparison, self-consistency,
Crocker, J., & Gallo, L. (1985, August). The self-enhancing effect of
and the concept of self. Journal ofPersonality and Social Psychology,
downward comparison. Paper presented at the annual meeting of the
16,148-156.
American Psychological Association, Los Angeles.
Crocker, J., & Schwartz, I. (1985). Prejudice and ingroup favoritism in a Myers, J. K., Lindenthal, J. J., & Pepper, M. P. (1972). Life events and
minimal intergroup situation: Effects of self-esteem. Personality and mental status: A longitudinal study. Journal of Health and Social
Social Psychology Bulletin, 11, 379-386. Behavior, 13, 398-406.
Crocker, J., Thompson, L. L., McGraw, K. M., & Ingerman, C. (1987). Nosanchuk, T. A., & Erickson, B. H. (1985). How high is up? Calibrat-
Downward comparison prejudice and evaluations of others: Effects ing social comparison in the real world. Journal of Personality and
of self-esteem and threat. Journal ofPersonality and Social Psychol- Social Psychology, 48, 624-634.
ogy, 52,907-916. Parducci, A. (1968). The relativism of absolute judgments. Scientific
Dakof, G. A. (1986). Psychological and social adaptation to Parkinson's American, 219, 84-90.
Disease. Unpublished doctoral dissertation, University of Califor- Rosenberg, M. (1965). Society and the adolescent self image. Princeton,
nia, Berkeley. NJ: Princeton University Press.
Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95, Salovey, P., & Rodin, J. (1984). Some antecedents and consequences of
542-575. social-comparison jealousy. Journal of Personality and Social Psy-
Festinger, L. (1954). A theory of social comparison processes. Human chology, 47, 780-792.
Relations, 7,117-140. Sarason, I. G, Johnson, J. H., & Siegel, J. M. (1978). Assessing the
Fiske, S. T., & Taylor, S. E. (1984). Social cognition. Reading, MA: Ad- impact of life changes: Development of the Life Experiences Survey.
dison-Wesley. Journal ofConsulting and Clinical Psychology, 46, 932-946.
 COMPARISON AND AFFECT 1249

Schachter, S. (1959). The psychology ofaffiliation. Stanford, CA: Stan- VanYperen, N. W, & Buunk, B. (1990). A longitudinal study of equity in
ford University Press. intimate relationships. European Journal of Social Psychology, 20,
Smith, R. H., & Insko, C. A. (1987). Social comparison choice during 287-309.
ability evaluation: The effects of comparison publicity, performance Vinokur, A., & Selzer, M. (1975). Desirable versus undesirable life
feedback, and self-esteem. Personality andSocial Psychology Bulle- events: Their relationship to stress and mental distress. Journal of
tin, 13,111-122. Personality and Social Psychology, 32, 329-337.
Stokols, D., Ohlig, W, & Resnick, S. M. (1978). Perception of residential Wheeler, L. (1966). Motivation as a determinant of upward compari-
crowding, classroom experiences, and student health. Human Ecol- son. Journal ofExperimental Social Psychology, Suppl. 1, 27-31.
ogy, 6, 33-57. Wheeler, L., Shaver, K. G, Jones, R. A., Goethals, G. R., Cooper, X,
Taylor, S. E., & Brown, J. (1988). Illusion and well-being: A social psy- Robinson, J. E., Gruder, C. L., & Butzine, K. W (1969). Factors deter-
chological perspective on mental health. Psychological Bulletin, 103, mining the choice of a comparison other. Journal of Experimental
193-210. Social Psychology, 5, 219-232.
Taylor, S. E., Falke, R. L., Shoptaw, S. J., & Lichtman, R. R. (1986). Wheeler, L, & Zuckerman, M. (1977). Commentary. In J. M. Suls &
Social support, support groups, and the cancer patient. Journal of R. L. Miller (Eds.), Social comparison processes: Theoretical and em-
Consulting and Clinical Psychology, 54, 608-615. pirical perspectives (pp. 335-357). Washington, DC: Hemisphere.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Taylor, S. E., Lichtman, R. R., & Wood, J. Y(l 984). Attributions, beliefs Wills, T. A. (1981). Downward comparison principles in social psychol-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

about control, and adjustment to breast cancer. Journal ofPersonal- ogy. Psychological Bulletin, 90, 245-271.
ity and Social Psychology, 46, 489-502. Wills, T. A. (1987). Downward comparison as a coping mechanism. In
Taylor, S. E., & Lobel, M. (1989). Social comparison activity under C. R. Snyder & C. Ford (Eds.), Coping with negative life events: Clini-
threat: Downward evaluation and upward contacts. Psychological cal and social-psychological perspectives (pp. 243-268). San Diego,
Review, 96, 569-575. CA: Academic Press.
Tesser, A. (1986). Some effects of self-evaluation maintenance on cog- Wilson, E. S. R., & Benner, L. A. (1971). The effects of self-esteem and
nition and action. In R. M. Sorrentino & E. T. Higgins (Eds.), Hand- situation upon comparison choices during ability evaluation. Soci-
book ofmotivation and cognition: foundations ofsocial behavior (pp. ometry, 34, 381-397.
435-464). New York: Guilford Press. Wood, J. V (1984). Social comparison in adjustment to breast cancer.
Tesser, A., Millar, M., & Moore, J. (1988). Some affective consequences Unpublished doctoral dissertation, University of California, Los
of social comparison and reflection processes: The pain and plea- Angeles.
sure of being close. Journal ofPersonality and Social Psychology, 54, Wood, J. Y (1989). Contemporary social comparison theory. Psycholog-
49-61. ical Bulletin, 106, 231-248.
Testa, M., & Major, B. (April 1988). Affective and behavioral conse- Wood, J. V, Taylor, S. E., & Lichtman, R. R. (1985). Social comparison
quences of social comparison. Paper presented at the annual meet- in adjustment to breast cancer. Journal of Personality and Social
ings of the Eastern Psychological Association, Buffalo, NY. Psychology, 49,1169-1183.
Thompson, S. C. (1981). Will it hurt less if I can control it? A complex
answer to a simple question. Psychological Bulletin, 90, 89-101. Received August 22,1989
Titus, S. L. (1980). A function of friendship: Social comparisons as a Revision received March 9,1990
frame of reference for marriage. Human Relations, 33, 409-431. Accepted March 17,1990 •