social comparison in adjustment to breast cancer

Journal of Personality and Social Psychology1985, Vol. 49, No. 5, 1169-1183

Copyright 1983 by the American Psychological Association, Inc.0022-3514/85/S00.75

Social Comparison in Adjustment to Breast Cancer

Joanne V. WoodState University of New York at Stony Brook

Shelley E. Taylor and Rosemary R. LichtmanUniversity of California, Los Angeles

We investigated four theoretical perspectives concerning the role of social comparison(Festinger, 1954) in coping with a threatening event in a sample of breast cancerpatients. According to the supercoper perspective, personal contact with comparisonothers is often unavailable to patients, and contact with media "supercopers"—fellow victims presented as adjusting very smoothly—may make patients feel in-adequate by comparison. According to the similarity perspective, patients selectcomparison targets who are similar to themselves because those comparisons shouldbe the most informative. The upward comparison perspective is predictive of com-parisons to relatively advantaged or superior individuals. The downward comparisonperspective leads to the prediction that under conditions of threat, individuals makecomparisons to people who are inferior or less fortunate in order to enhance theirself-esteem. We interviewed 78 breast cancer patients, and results of both closed-ended questions and spontaneously offered comparisons yielded a preponderanceof downward comparisons. The results point to the value of using naturalistic meth-ods for studying comparisons, and suggest a more active and cognitive role for socialcomparison than is usually portrayed.

Nearly all persons are confronted with se-rious illness in their lifetimes, either by be-coming ill themselves or by being faced withthe illness of someone they love. The difficultieswrought by serious illness include pain anddisability, disruption of one's life style, intensefears, and strains in close relationships (Cohen

This article is based on the first author's doctoral dis-sertation, which was submitted to the University of Cali-fornia, Los Angeles (UCLA). She is grateful to Barry Col-lins, Chris Dunkel-Schetter, Oscar Grusky, Connie Ham-men, David Wellisch, and especially John Michela for theiradvice and encouragement. We all thank Barbara Futter-man and Patricia Loftus for their aid in data collection,Carol Wixom and Marlene Lukaszewski for their assistancein coding data, and Avrum Bluming, Robert Leibowitz,and Gary Dosik for permitting us to interview their pa-tients. We are also grateful for the helpful contributionsof two anonymous reviewers.

The research was supported by research funds fromUCLA to all three authors, research funds from the Na-tional Institute of Mental Health (NIMH) to the secondauthor (MH 34167), and by a Research Scientist Devel-opment Award (MH 00311) to the second author. The firstauthor was also supported by an NIMH training grant andby a National Institutes of Health Biomedical ResearchSupport Grant (SO7 RR0706720) at Stony Brook.

Requests for reprints should be sent to Joanne V. Wood,Department of Psychology, State University of New Yorkat Stony Brook, Stony Brook, New York 11794.

& Lazarus, 1979; Wortman & Dunkel-Schet-ter, 1979). How do people adjust to thesestressors? Social comparison (Festinger, 1954)may be important in adjustment through itstwo main functions: self-evaluation and self-enhancement.

Self-Evaluation and AdjustmentAccording to Festinger's (1954) theory, self-

evaluation is the primary function of socialcomparison. He proposed that people have adrive to evaluate their opinions and abilities,and that they often do so by comparing them-selves with others. Schachter (1959) extendedthe theory by proposing that individuals seekto clarify their emotional reactions in stressfulsituations, and this position has received lab-oratory support (Cottrell & Epley, 1977). Somewriters have suggested that patients seek toevaluate their emotional reactions to illness(Mechanic, 1977; Wortman & Dunkel-Schet-ter, 1979). Wortman and Dunkel-Schetter, forexample, wrote that

A person diagnosed with cancer is likely to be highly fearfuland uncertain. The intensity of their feelings and anxietiesmay lead many patients to worry that they are coping poorlyor losing their grip on reality. They experience a need to. . . learn whether their reactions are reasonable and nor-mal, (p. 124)

1169

1170 J. WOOD, S. TAYLOR, AND R. LICHTMAN

Whether social comparison is important to thisprocess of self-evaluation has not been studieddirectly. However, people grappling with criseswant to express their feelings and to discussthem with others (Dunkel-Schetter & Wort-man, 1982; Silver & Wortman, 1980). Onefunction of such discussions may be to helpindividuals evaluate their reactions by com-paring them with those of fellow victims.

Self-Enhancement and Adjustment

The desire for an accurate self-evaluationmay clash at times with the desire to obtain afavorable self-evaluation (cf. Cruder, 1977;Singer, 1966), and self-enhancement has beenidentified as a motive for social comparison(e.g., Hakmiller, 1966). The results of labo-ratory studies suggest that a critical determi-nant of which motive is paramount seems tobe a threat to self-esteem; under threat, indi-viduals often opt for a self-enhancing strategy(Brickman & Bulman, 1977; Cruder, 1977).For example, they may avoid comparisons withothers who appear to be superior.

Serious illness can pose great threats to self-esteem because it can bring many changes thatare critical to one's identity: body image, oc-cupation, valued activities, and close relation-ships (Cohen & Lazarus, 1979; Wortman &Dunkel-Schetter, 1979). Patients may be mo-tivated, therefore, to enhance their self-esteemin a number of areas (Pearlin & Schooler, 1978;Taylor, 1983). Their self-enhancement goalsmay well be served by social comparison.

In summary, two hypothesized functions ofsocial comparison may be involved in adjust-ment to illness: Social comparison may be usedto evaluate one's emotional reactions and maybe used for self-enhancement. We examinedfour theoretical perspectives, each of whichconcerns one or both of the two motives forsocial comparison. These perspectives arecalled the supercoper perspective, the similarityperspective, the upward comparison perspec-tive, and the downward comparison perspec-tive.

The Supercoper Perspective

The supercoper perspective concerns howsocial comparison can affect a victim's self-evaluation of her adjustment. This perspective

is drawn from Taylor and Levin's (1976) reviewof the breast cancer literature, in which theysuggested that a range of comparison others isunavailable to many patients. Many women,for example, have no personal friends who havebeen through the same experience. One sourceof comparisons that may be more widelyavailable is the media. However, Taylor andLevin (1976) observed that media coverage isoften biased toward those who are adjustingvery successfully. With respect to breast cancer,for example, well-known patients such asMarvella Bayh and Shirley Temple Black areoften presented as being able to pick up theirlives easily where they left off before their sur-gery. This media bias may not be unique tocancer; TV news coverage of many tragicevents often seems to focus on victims whoadjust well (cf. Wortman & Dunkel-Schetter,1979). An unfortunate consequence of thismedia bias is that victims who are exposed tothese "supercoper" models but who themselvesare experiencing bouts of depression and self-doubt may feel inadequate.

Thus the supercoper perspective leads to twopredictions: (a) Comparison others are gen-erally unavailable to breast cancer patients,and, consequently, patients compare them-selves with media figures; and (b) because themedia disproportionately feature supercopers,comparisons with media figures make womenfeel worse about themselves. Whereas the su-percoper perspective concerns the effects ofcomparisons, the remaining three perspectivesconcern selection of comparison others.

The Similarity Perspective

Like the supercoper perspective, the simi-larity perspective is concerned with self-eval-uation of adjustment. Festinger (1954) main-tained that in order for one to make an ac-curate self-evaluation, one must compareoneself with similar others. However, this sim-ilarity hypothesis is complicated by debate overwhat basis of similarity is important. Accord-ing to some authors, it is similarity on the di-mension under evaluation (see Goethals &Darley, 1977). In the context of self-evaluationof adjustment, such dimension-specific com-parisons would involve comparisons with oth-ers who are adjusting similarly.

Other authors have contended that what

COMPARISON IN ADJUSTMENT 1171

Festinger (1954) meant was similarity on at-tributes related to the dimension under eval-uation (e.g., Goethals & Darley, 1977), or whathas been termed related attributes similarity(Wheeler & Zuckerman, 1977). Zanna, Goe-thals, and Hill (1975) provided this exampleto illustrate the distinction:

A swimmer evaluating his [or her] ability would prefercomparing with another person who would have a similartime given his [or her] age, experience, and recent practice[all attributes related to swimming ability], rather thansimply a person with a similar time [dimension-specificsimilarity], (p. 87)

An informative comparison strategy, then, isto select a person who is similar on relateddimensions, and then compare with that per-son in terms of the dimension under evalua-tion. Attributes that breast cancer patientsconsider to be related to adjustment—andtherefore important to comparison selection—may include type of surgery, prognosis, andage. According to the similarity hypothesis,then, victims will compare themselves withwomen who are similar on the specific dimen-sion in question or on attributes related to thatdimension.

The Upward Comparison Perspective

Many laboratory studies have indicated thatsubjects often compare themselves with some-one in the desirable direction—that is, some-one who possesses a lot of a favorable attributeor very little of a negative attribute (Cruder,1977). Upward comparison may serve eitherthe purpose of self-evaluation or of self-en-hancement. Breast cancer patients may, forexample, attempt to measure their adjustmentagainst that of someone who is better off, orthey may feel better by identifying with her (cf.Brickman & Bulman, 1977). For either pur-pose, then, victims will compare themselvesto other patients who are better off than theyare, according to the upward comparison per-spective.

The Downward Comparison Perspective

As discussed earlier, concerns for an honestself-evaluation are at times outweighed by adesire to preserve one's self-esteem (cf. Singer,1966). Several studies have shown that com-parisons with superior others are avoided when

one's self-esteem has been threatened (e.g.,Wilson & Benner, 1971). Moreover, under suchconditions people may compare themselveswith others who are inferior or less fortunate.Such "downward comparisons" have been re-ported in the experimental literature (e.g.,Hakmiller, 1966; Wilson & Benner, 1971), andsuggestive evidence has appeared in the copingliterature (Pearlin & Schooler, 1978). Wills(1981) offered the first formal theoreticalstatement of downward comparison. He re-viewed a large literature to support his thesisthat under victimizing circumstances peopleoften attempt to preserve their self-esteemthrough downward comparison.

The downward comparison idea is consis-tent with a more general conceptualization ofvictims' reactions that was recently proposed(Taylor, Wood, & Lichtman, 1983). When in-dividuals become the victims of crime, illness,or natural disasters, Taylor et al. (1983) argued,they often seek to minimize their sense of vic-timization. Victims may use strategies of se-lective evaluation that attempt to limit theirown and others' perceptions of how bad theirsituations are; downward comparison is onesuch strategy. Taylor et al. (1983) identifiedfour additional strategies of selective evalua-tion: creating hypothetical, worse worlds (e.g.,"It could have been worse"); construing benefitfrom the victimizing event (e.g., "I am a stron-ger person now"); manufacturing normativestandards of adjustment, compared with whichone's adjustment is good (e.g., "So many arejust devastated, but I'm certainly not"); andselectively focusing on dimensions that makeone appear advantaged, namely, "dimensionalcomparisons" (e.g., "At least I'm married—itmust be so hard for these single women"). Ofthese four strategies, comparisons with nor-mative standards and dimensional compari-sons most clearly involve social comparisonprocesses. Hence they are also examined inthis study. According to the downward com-parison perspective, victims will favor down-ward comparisons and these related selectiveevaluation strategies.

Method

Recruitment of Subjects

Subjects were obtained through a three-physician privateoncology practice in Los Angeles. Thirty patients were


screened out because of severity of illness or geographicinaccessibility. A letter describing the study was sent to theremaining 179 patients by the physicians' nurse, and 87interested patients contacted the researchers, which yieldeda response rate of 49%. Nine of these women eventuallydid not participate because of logistical problems.

SampleThe final sample included 78 women ranging in age from

29 to 78 (mean age was 53). Seventy-one percent weremarried. The mean level of education attained was oneyear of college. Overall, the economic distribution suggestsa skew toward the middle and upper socioeconomic classes,and the sample had a somewhat disproportionate Jewishrepresentation: Protestant (46%), Catholic (15%), Jewish(31%), other (4%), and no religion (4%).

All but 3 of the women had been treated surgically fortheir breast cancer. Thirty-five percent of those (26) hadhad a lumpectomy (removal of the malignant lump andsome supportive tissue), 3% (2) had had a simple mastec-tomy (removal of the breast), 39% (29) had had a modifiedradical mastectomy (removal of the breast and some ad-jacent lymph nodes), 12% (9) had had a Halsted radicalmastectomy (removal of the breast, adjacent lymph nodes,and part of the pectoral muscles), and 12% (9) had hadsurgery on both breasts. Thirty-one percent of the patientswere initially diagnosed as having Stage I cancers, 55% hadStage II cancers, and 14% had distant sites of metastases.Length of time since surgery ranged from 2 months to 16years; the median was 25 '/2 months (For additional infor-mation about the sample, see Taylor, Lichtman, & Wood,1984).

Procedure

Respondents were telephoned and an interview was ar-ranged, usually in the home. The interviewers were theauthors and two other experienced female interviewers.The standardized, tape-recorded interview typically lastedbetween 1.5 and 2 hr and it covered (in order) demographicdata, the woman's cancer experience and its treatment,attributions for cancer and beliefs about its controllability,life changes since cancer, changes in close relationships,specific fears about the cancer (e.g., death, finances), generalemotional reactions (fear, anxiety, depression, and anger),social comparison processes, and compliance with medicalregimen. Because the social comparison section of the in-terview was most critical to the present study, it is describedafter the remaining measures. The specific form of anyother questions used in the analyses is covered in the Resultssection.

Questionnaire. A questionnaire was left with respon-dents, and the return rate was 90%. The only questionnairemeasure involved in this study is the Locke-Wallace Scaleof Marital Adjustment (Locke & Wallace, 1959).

Interviewer and physician ratings. Independent ratingsof psychological adjustment to the illness were obtainedfrom the interviewer and the physician on the Global Ad-justment to Illness Scale (GAIS; Derogatis, 1975). Usinginterview and chart materials, interviewers also made aphysical state rating for each patient. This 7-point scaleranged from 1 (obviously deteriorating from metastatic

cancer) to 7 (prognosis good, e.g., a small tumor, no nodalinvolvement, symptom-free for at least 2 years).

Social Comparison Section of Interview

The social comparison questions tapped three generalcategories: (a) contacts with potential comparison others,(b) impressions about other patients' coping, and (c) com-parisons made. First, several questions concerned the con-tacts the patients had with various media and social sources.Patients were asked how much contact they had had withcancer-related books, TV shows, newspaper and magazinearticles (these were coded on 4-point scales, from 1 = noneto 4 = a lot); other media sources (1 = yes, 2 = no); otherwomen and other breast cancer patients (1 = none, 4 =Quite a few); support groups (1 = no, 2 = yes, went once,3 = continued to go); and Reach-to-Recovery volunteers(1 = yes, 2 = no). In an effort to obtain an index of contactsthat would be more reliable than the individual contactitems, we created a contact composite through principalcomponents analyses. This media composite consisted ofbooks, TV shows, and newspaper and magazine articles.Its reliability, as measured via Cronbach's alpha, was .65.

Next, patients were asked what impressions they hadformed of other women's coping as a result of their contactsand were asked, "Have you ever thought about how youhave coped in comparison with any of these impressions?(1 = yes, 2 = no); then, "In general, have you thought youwere coping much better than other women with breastcancer, coping somewhat better, coping about the same,coping somewhat worse, or coping much worse?" Responseswere coded on a 6-point scale from 1 = coping much betterto 6 = coping much worse (the response "better than some,worse than others" was coded as 4). This scale is called thecoping-in-comparison rating.

Free-Response Social Comparison Measures

Once data were collected and the interviews transcribed,it became clear that phenomena relevant to social com-parison had occurred in instances other than the socialcomparison section of the interview. Throughout their in-terviews, most women spontaneously had made compar-isons with other people. These comparative remarks offeredus an opportunity to examine the theoretical perspectivesthrough a measurement method different from the cus-tomary structured, closed-ended comparison questions.Their spontaneous quality suggests that the free-responsecomparisons occur naturally and may therefore be morecentral to the patients' experience than the comparisonselicited by investigator-designed questions.

Both descriptive and theoretical considerations guidedthe creation of the coding system for these free responses.In order to describe them fully, comparisons in differentcontent domains—such as psychological adjustment andphysical status—were kept separate. We also derived cat-egories to address certain theoretical questions, as describedlater. To compute interrater agreement, the first authorand a second coder each coded 24 randomly selected in-terview transcripts. Agreements were scored when partic-ular quotations were selected and coded identically. All


Table 1Free-Response Measures: Descriptions of Others' Adjustment

Category

Good coper

Poor coper

Definition

Description ofpatient reactingor adjustingwell

Description ofpatient reactingor adjustingpoorly

Example

She was very calm andcool and reassuring,and told me therewas no painafterwards.

I talked to the nurse,who had it just twoweeks after I did,and she says, "I'mnot making a goodadjustment at all.I'm having aterrible time."

No.comparisons

made

012346

012356

Frequency

50173111

31296421

%"

68.523.3

4.11.41.41.4

42.539.78.25.52.71.4

" Percentages were computed from 73 interviews; 5 others were not transcribed because of equipment failure.

interrater agreements were computed as percentage ofagreement on occurrences only.1

Comparisons Relevant to theSupercoper Perspective

In order to examine the supercoper predictions, it isimportant to determine whether direct, personal contactsare available as comparison targets when one evaluates one'spsychological adjustment. Hence the transcripts were codedfor the presence of two responses, which are presented inTable 1. Descriptions of the emotional reactions or psy-chological adjustment of a seriously ill other with whomthe patient had direct, personal contact were divided intotwo categories: good copers and poor copers; interrateragreements were .80 and 1.00, respectively.

Comparisons Relevant to theSimilarity Perspective

Because the similarity perspective concerns self-evalu-ation of psychological adjustment, several categories ofcomparisons involving relative adjustment were coded.

Others cope better, others cope worse, and others copethe same. Each quotation that involved a comparison onthe dimension of psychological adjustment was coded asto whether the respondent explicitly stated that she wasadjusting more poorly than, similarly to, or better thananother. An example of others cope better is as follows:

The girl across the street went in and had her breastremoved.. . .A week later she was out working, cheerful,and seemed just fine. And look at me after all this time.

Others cope the same:

There was a movie on about a young girl and how shecoped with cancer. . . . I remember feeling the sameway that she seemed to feel.

Others cope worse:

I have never been like some of those people who havecancer and they feel well, this is it, they can't do anything,they can't go anywhere. . . . I just kept right on going.

In keeping with the nature of free-response comparisons,these three categories were not drawn from the interviewquestions in which we explicitly asked about comparisons.2

1 Only a brief description of each free-response com-parison measure is provided here. One may obtain moredetailed information about coding by writing to the firstauthor.

2 It is possible that a given statement could meet thecriteria for one of the categories mentioned earlier—goodand poor coper—or one of these categories, if, for example,a respondent both described a successfully adjusting other(good coper) and explicitly said that the other was adjustingbetter than she was (others cope better). Both types of cat-egories are necessary because they have different meanings.For example, it is one thing to describe another woman asadjusting well, and it is quite another to add that she isadjusting better than oneself. A given statement could bean others cope better comparison and not a good copercomparison, and vice versa (e.g., one could describe anotheras a good coper and state that her own coping was evenmore successful, i.e., others cope worse). Throughout ourstudy, the coding of a statement in any given category didnot constrain how it was coded in any other category (e.g.,no category was a subset of any other).


The interrater agreements for others cope the same andothers cope better were both .50, and for others cope worseit was .86.3

Related attributes. Related attributes were comparisonsthat involved the respondent's acknowledgment that ad-justment may depend on such factors as marital status,age, physical status, and so on. The comment had to bedescriptive of another person's psychological adjustmentor emotional reactions, and had to strongly imply that acertain factor (such as age, marital status) influenced theother's adjustment. One woman, for example, said,

I think maybe I'm coping a little better, but these otherwomen were divorced. That makes a big difference inhow you cope. I have a lot of stability and a lot of love,and that helps me cope.

The interrater agreement was .83.4

Comparisons Relevant to the Upward andDownward Comparison Perspectives

The upward and downward comparison predictions maybe tested by means of using some of the measures alreadydescribed, such as the ratio of good copers to poor copersand the ratio of others cope better to others cope worse.Additional measures were coded also.5

Manufacturing normative standards of adjustment. Inone comparative strategy, the victim contrasts her own ad-justment with a standard that seems to be based less onpersonal contact with fellow victims than on beliefs abouthypothetical others (Taylor et al., 1983). She "manufacturesa normative standard of adjustment," compared with whichher adjustment is very good. An example is as follows:

I have heard, second hand, that some of them, manyyears down the road, are still not over i t . . . . There arewomen who don't ever reach this point [that I have].

To help ensure that the normative standards were not basedon direct experience, we did not include statements thatmentioned specific, known others. The interrater agreementwas .80.

The measures discussed thus far involve comparisonsmade in the domain of psychological adjustment becausethe supercoper and similarity perspectives both involve self-evaluation of adjustment. However, the upward and down-ward comparison perspectives also involve self-enhance-ment, and self-enhancing comparisons may be made inmany domains other than adjustment (e.g., physical andmarital status). Hence other types of free-responses werecoded as well.

Physical comparisons. Descriptions of specific othercancer patients' physical condition were grouped into threecategories, as described in Table 2. Upward physical anddownward physical comparisons involved patients whowere physically better off and physically worse off than theinterviewee, respectively. Although generally downwardcomparisons are viewed as enabling one to feel better aboutone's situation, some comparisons in the downward di-rection were clearly threatening. Take the following ex-ample:

His wife had breast cancer, eventually cancer of the bonemarrow, and I saw the wheelchairs.... I knew she wasin the hospital and lost her hair. And I had my breast

surgery by this time, and I'm suddenly thinking oh, myGod, is this going to be me, this kind of future?

Such threatening physical comparisons were coded sepa-rately from other downward comparisons. Interrateragreements for upward, downward, and threatening phys-ical comparisons were .73, .80, and .86, respectively.

Situational comparisons. Some free-response compar-isons involved other cancer patients whose life situations,according to the respondent, were clearly worse or betterthan her own. Situational comparisons did not includephysical state or psychological adjustment, which werecoded elsewhere. Rather, they involved some other aspectof the target person's life, such as her family. An exampleof an upward situational comparison is as follows:

I was the only one that was working, everyone else hadhusbands to support them and take care of them.. . .I had to get out and get going when the rest of themcould lay around.

A downward situational example is as follows:

I have heard stories where men walked out on their wives,couldn't accept it. My friend, her husband couldn't dis-cuss it. He didn't want to hear anything about it.

The interrater agreement for upward situational re-sponses was 1.00, and for downward situational responsesit was .75.

Dimensional comparisons. Most of the categoriesidentified so far involve selecting a comparison target whois relatively disadvantaged or advantaged. Taylor (1983)and Taylor et al. (1983), however, have identified a com-parison strategy in which a comparison target is not se-lected. Rather, the comparer selects a comparison dimen-sion and evaluates her own standing on the dimension, inrelation to an alternative. For example, one breast cancerpatient said,

There are days when I look in the mirror and I am upsetwith the scar under my arm and I think to myself, "Youare upset with that; how would you feel with a mastec-tomy scar?"

Note that the object of comparison is not a disadvantagedperson, but a dimension of comparison (in this case, sizeof scar) on which the comparer is relatively better off.Downward dimensional and upward dimensional compar-isons involved dimensions on which the respondent was

3 The low interrater agreements for others cope the sameand others cope better is due to their low frequencies; theyeach occurred only once or twice in the 24 transcriptsfrom which agreement was computed.

4 As one can see in the example, both a related attributeand an others cope worse could be coded from these state-ments. It is also clear, however, that these categories tapinto different meanings and are based on different phrasesin the passage. The related attributes category has similarpotential overlap with some categories to follow (e.g., sit-uational comparisons), but in no case are the categoriesredundant in meaning.

5 All remaining categories were mutually exclusive (i.e.,a given comparison could not be coded into more thanone category).


advantaged or disadvantage*!, respectively. An example ofan upward dimensional comparison is as follows:

They accept the woman like this, like I am now, whenthey're already married for years, or they already loveher.. . . But I was single. . . . I don't have a man, andthis is what really hurt me.

In order to be coded as a dimensional comparison, astatement had to be an explicitly made contrast betweenthe patient's situation and that of others, and it could notinvolve specific individuals, so that the focus was on thedimension of comparison rather than on comparison tar-gets. Interrater agreements were .89 for downward and 1.00for upward dimensional comparisons.

Results

Representativeness of Sample

To examine the possibility of sample selec-tivity, we compared the 87 patients who vol-unteered to participate with the 92 who hadnot. Results of t tests revealed no significantdifferences between the two groups on demo-graphic, disease, or treatment characteristics(see Taylor et al., 1984, for details) and on phy-sicians' GAIS ratings. Thus it appears that thewomen who chose to participate did not differin terms of physical status and psychologicaladjustment from those who did not. In addi-tion, the respondents' GAIS scores (M = 82)fell within the range identified by other studiesof cancer patients (Derogatis, Abeloff, & Mel-isaratos, 1979; J. L. Michela, personal com-munication, March, 1984).

Next we turn to the four theoretical per-spectives. The results of the free-response cod-ing are presented in two ways. First we presentthe frequencies of each category, and the per-centage of the sample of 73 who made eachcomparison.6 Second, chi-square analyses arereported when it is appropriate to comparethe frequencies of categories (e.g., upward sit-uational with downward situational). Ratherthan to compare the numbers of each com-parison totaled across the sample, however, thechi-squares were used to compare the numberof respondents who made each comparison.The rationale for this decision is that the totalnumbers would have included multiple com-parisons made by single respondents. Hencethe number of respondents making each com-parison seems to be a more meaningful andconservative index of the use of comparisons.

The Supercoper PerspectiveAccording to the supercoper perspective,

comparison others are not widely available topatients, and consequently patients are forcedto compare themselves with media figures.Because the media disproportionately featuresupercopers, these comparisons may lead awoman to feel inadequate.

The results indicate that the first assumptionof the supercoper perspective, that comparisonothers are not widely available to patients, isinaccurate. As one can see in Table 3, theoverall degree of contact with both media andsocial sources was high. Most women had readat least one book on cancer, and over 90% saidthey had read newspaper or magazine articles.Eighty-five percent of the respondents said thatthey had discussed their cancer with otherwomen, and over 66% said that they had spo-ken with "a few" or "quite a few" other breastcancer patients. Moreover, 49 women (67%)spontaneously described another patient's ad-justment or emotional reactions (either goodcopers or poor copers, as described in Table1). Taken together, these findings suggest thatinformation about others' adjustment to can-cer is more widely available than is assumedin the supercoper perspective, and that victimsare not dependent on the media for compar-ison targets.

Despite this fact, our results supported thesecond prediction of the supercoper perspec-tive, that the more contact a woman has withthe media, the worse she thinks she is copingcompared with other women: The mediacomposite was negatively correlated withthe coping-in-comparison rating (r = -.26,p < .03).7

The Similarity Perspective

According to the similarity perspective, in-dividuals compare themselves with those who

6 The relative frequency percentages are based on 73interview transcripts; 5 of the original 78 interviews werenot transcribed because of audiotape equipment failure.Several other transcripts (9, or 11.5%) were only partialtranscripts; that is, a portion of the interview was missing.These transcripts were included in the frequencies, despitethe fact that their inclusion undoubtedly lowers the relativefrequencies of some categories. Therefore, the frequenciesare probably often conservative estimates of the actualnumber of comparisons made.

7 All correlational tests were one-tailed.


Table 2Free-Response Measures: Physical Comparisons

Category

Upwardphysicalcomparisons

Downwardphysicalcomparisons

Threateningphysicalcomparisons

Definition

Description ofpatient who isphysicallybetter off

Description ofpatient who isphysicallyworse off

Description offeeling scaredor threatenedby physicallydisadvantagedothers

Example

She had a tumor the sizeof the end of herfinger, which I wouldhave been glad totrade her even.

At first [the scar] wasgross. . . . Now I don'tthink it's so bad,especially after you'veseen my friend; shejust had two radiationimplants put in.

I went to a CancerSociety meeting oneevening . . . and Irealized how manywomen had bothbreasts removed.Until that time, it hadnever occurred to me.And I've never goneback to anothermeeting, because Ihad nightmares thatnight.

No.comparisons

made

012

012345

0123

Frequency

7021

341810821

541441

%a

95.92.71.4

46.624.713.710.92.71.4

74.019.25.51.4

" Percentages were computed from 73 interviews; 5 others were not transcribed because of equipment failure.

are similar on specific dimensions of evalua-tion or on related attributes. However, resultssuggest that comparisons with similar othersare relatively infrequent.

Dimension-specific similarity. Only 12(16.4%) of the women stated that another per-son (or persons) was adjusting in a similarmanner (others cope the same). Although fewer(6, or 8.2%) said others were adjusting better(others cope better), the majority of the women(44, or 60.3%) indicated that others were cop-ing more poorly (others cope worse). Of the 38respondents who made any or all of these threecomparisons, 3 women most often made"others cope better" comparisons (i.e., theymade more comparisons with others cope bet-ter than with others cope the same and withothers cope worse); 3 most often made otherscope the same comparisons; and 32 most oftenmade others cope worse comparisons. The chi-square analysis in which we compared thesefrequencies was highly significant, x2(2, N =38) = 44.26, p<. 001.

On the coping-in-comparison rating, womenwere specifically asked to make a judgment interms of dimension-specific similarity. Nearlyhalf (28, or 46.6%) of the 60 who respondedrated their adjustment as "much better thanother women with breast cancer"; anotherthird (20, or 33.3%) rated it as "somewhat bet-ter"; 8 (13.3%) said they were coping "aboutthe same"; 2 (3.3%) said, "better than some,worse than others"; no one said, "somewhatworse"; and 2 (3.3%) said they were coping"much worse than other women with breastcancer."8 A goodness-of-fit test in which we

8 Thirteen women declined to rate their comparison onthe 6-point scale, usually saying that they "don't know"or that they had no one with whom to compare. Resultsof a t test suggested that those women who did not ratetheir comparison did not differ from those who did, interms of their emotional adjustment, t(l 1) = .51, p < .62;emotional adjustment was measured via a factor score thatconsisted of both standarized instruments and interviewquestions; see Taylor, Lichtman, and Wood (1984) for de-tails.


Table 3Breast Cancer Patients' Mediaand Social Contacts

Contact source Frequency

Media sources"

BooksNoneOneSomeA lot

TV showsNoneOneSomeA lot

Newspaper andmagazine articles

NoneA fewQuite a fewEverything I can read

Other media"NoYes

271723

7

1525276

7302710

4823

36.523.031.19.5

20.534.237.0

8.2

9.540.536.513.5

67.632.4

Social sources

Other womenNoneYes, 1 or 2Yes, a fewQuite a few

Other breast cancerpatients

NoneYes, 1 or 2Yes, a fewQuite a few

Support groupsNoWent onceContinued to attend

Reach-to-Recoveryvolunteer

NoYes

116

544

141148

2

5220

2

3833

14.78.0

72.05.3

18.714.764.0

2.7

70.327.02.7

53.546.5

' The media items specifically inquired about cancer-relatedmaterial.b "Other media" referred to pamphlets, radio programs,and so on.

compared this scale's distribution with a nor-mal distribution indicated that they were sig-nificantly different, x2(5, N = 60) = 568.40,p < .001. These results suggest that these pa-tients do not typically compare with otherswho are close to themselves on the adjustmentcontinuum.

Related attributes similarity. Over 60% ofthe 73 respondents mentioned "related attri-butes" in comparisons of psychological ad-justment. Thirty-five women (47.9%) madeone such comparison, and 9 (12.3%) made two.Several categories of related attributes wereparticularly common. A number of women(13, or 17.8%) mentioned the importance ofmarital status or support of loved ones. Otherfrequent related attributes were age (men-tioned by 11 women, or 15.1 %) type of surgery(9, or 12.3%) and prognosis (9, or 12.3%).

Overall, respondents frequently made com-parisons in terms of proximity along the di-mension of psychological adjustment (dimen-sion-specific similarity) and in terms of attri-butes related to adjustment (related attributessimilarity). However, close similarity does notseem to be critical to either type of compari-son. Women rarely compared themselves withothers who were similar on the dimension ofadjustment, and they usually described relatedattributes on which they were very differentfrom comparison others, as the example pro-vided in the Method section illustrates.

Upward Versus Downward ComparisonPerspectives

Two results already reported strongly suggestthat the women were making many downwardcomparisons: Over 60% of respondents saidthat another patient was coping less well thanshe was (others cope worse), and when ratingtheir adjustment relative to other women withbreast cancer (coping-in-comparison rating),80% said that they adjusted at least "somewhatbetter." We also examined additional free-re-sponse comparisons to test the upward anddownward perspectives. With a few exceptionsthat we will note, in all chi-square analyses inwhich we compared upward comparisons todownward comparisons, we compared threefrequencies: (a) the number of respondentswho made more upward than downward com-parisons; (b) the number who made an equalnumber of upward and downward compari-sons; and (c) the number who made moredownward than upward comparisons. The chi-squares are goodness-of-fit tests in which wecompared the observed distribution with adistribution with equal frequencies in the threecategories.


We first examined respondents' descriptionsof others' coping. As shown in Table 1, about31% of the sample gave an example of a goodcoper, whereas about 57% gave an example ofa poor coper. The frequencies of respondentswho described more good copers than poorcopers (n = 11), who described an equal num-ber of both (n = 7), and who described morepoor than good copers (n = 31) were signifi-cantly different, x2(2, AT = 49) = 20.2, p <.001. Similarly, for manufacturing normativestandards of adjustment, 20 women (27.4%)made a comparison with a downward nor-mative standard; that is, they described a hy-pothetical standard in comparison with whichtheir adjustment was exceptionally good. Onlyone woman (1.4%), in contrast, described mostother women as adjusting better than she had.The difference between these two frequenciesis highly significant, x20, N = 21) = 17.2,p < .001. Clearly, then, several results suggestthat patients were making many downwardcomparisons in the domain of psychologicaladjustment.

Physical comparisons. We compared thenumber of upward physical and downwardphysical comparisons and found a similar re-sult. As shown in Table 2, only 4.1% of thesample mentioned another person who wasphysically better off than they were, whereasover half of the women described another per-son who was comparatively worse off. Of the40 women who made either of these compar-isons, 1 made more upward than downwardcomparisons, 1 made an equal number of both,and 38 made more downward than upwardcomparisons, x2(2, N = 40) = 68.45, p < .001.

However, some comparisons with physicallyworse off others may be threatening. A sizablenumber of women (26%) described threateningphysical comparisons—that is, feeling threat-ened by comparing with physically disadvan-taged others. Although these comparisons werevoiced only about half as frequently as otherdownward physical comparisons, they illus-trate that downward comparisons are not al-ways self-enhancing.

Situational comparisons. Situational com-parisons involve specific other cancer patientswhose life circumstances (e.g., marital satis-faction and age) were better or worse than therespondent's. Fourteen women (19.2%) madeat least one downward Situational comparison

(1 woman made two), whereas only 2 (2.7%)made an upward Situational comparison.These two figures (no women made both com-parisons) were significantly different, \2(\,N =16) = 9.0, p = .003.

Dimensional comparisons. In dimensionalcomparisons, the respondent emphasizes a di-mension rather than a specific comparisonperson. Downward dimensional comparisonswere found in 47 (64.4%) of the transcripts.Thirty-seven women (50.7% of the sample)made one such comparison, 8(11.0%) madetwo, and 2 women (2.7%) made three. Upwarddimensional comparisons were made muchless frequently: only 7 women (9.6%) madethese. Of the 48 women who made dimensionalcomparisons, 1 made more upward thandownward, 3 made an equal number, and 44made more downward than upward, \\2, N =48) = 73.62, p<. 001.

The most frequently occurring downwarddimensional comparison (over 20% of them)involved type of surgery, such as the followingexample:I had just a comparatively small amount of surgery oa thebreast, and I was so miserable, because it was so painful.How awful it must be for women who have had a mastec-tomy. . . . I just can't imagine it, it would seem to be sodifficult.

Nineteen percent of the downward dimen-sional comparisons involved older womencomparing themselves favorably with youngerwomen. Also frequent were statements madeby mastectomy patients comparing their breastloss with greater disfigurements (15% of thedownward dimensional comparisons), andcomparisons involving chemotherapy (15%)marital status (10%), and other illnesses (8.5%).

In summary, these data from the domainsof psychological adjustment, physical status,and life situations offer strong support for thedownward comparison perspective: Down-ward comparisons were made much more fre-quently than were upward comparisons.

Totals of upward and downward compari-sons. But what if this discrepancy betweendownward and upward comparisons is largelydue to a small group of patients who makedownward comparisons in every category? Wecreated two summary comparison indices—upward and downward—by collapsing acrossfree-response categories. The upward com-parison index was the sum of the respondent's


good coper, manufacturing normative stan-dards of adjustment, upward physical, upwardsituational, and upward dimensional compar-isons; the downward comparison index was thesum of the downward counterparts of each ofthese measures. (Others cope better and otherscope worse were not included, so as not tooverrepresent the adjustment domain). Resultsof a chi-square analysis in which we used theseindices support the conclusion that downwardcomparisons are prevalent across the entiresample. In this test we compared four groups:the numbers of respondents who (a) gave nofree-response comparisons at all (n = 4), (b)made more upward than downward compar-isons (n = 1), (c) made an equal number ofboth (n = 5), and (d) made more downwardthan upward comparisons (n = 63). The chi-square was highly significant, x2(3, N = 73) =146.78, p < .001. These data, then, offer re-sounding support for the downward compar-ison perspective.

Predictors of Downward Comparison

What factors contribute to making down-ward comparisons? One factor may be threat(Taylor et al., 1983; Wills, 1981). Several vari-ables that may be indicators of stress or threatwere correlated with the index of downwardcomparison described earlier.

Cancer- and treatment-related sources ofthreat. Type of surgery should be an indicatorof threat; those women who have had moreradical surgeries have presumably undergonemore threat to their body image. Type of sur-gery was broken down into three groups (lum-pectomy, simple and modified mastectomy,and Halsted radical), but it was not signifi-cantly related to the downward comparisonindex (r = —. 13). Although poor prognoses areclearly threatening, prognosis (as measured bythe physical state rating) was also unassociatedwith downward comparisons (r = —.01).

Social-situational sources of threat. Socialsupport may help the patient feel less threat-ened, and hence should be negatively corre-lated with downward comparison. However,scores on the Locke-Wallace Marital Adjust-ment Scale (Locke & Wallace, 1959) and thepatient's ratings of satisfaction, communica-tion, and perceived support in her relationshipwith her significant other (each rated on 4-

point scales) were uncorrelated with thedownward comparison index (all rs < .18).Scores on one index, support from family,suggested that the less the woman feels sup-ported by her family, the more she makesdownward comparisons (r = -.37, p = .001).However, the overall results for the cancer-re-lated and social support sources of threat donot support the view that threat leads todownward comparison.

When do patients make downward compar-isons? Downward comparisons may be seenas preliminary attempts at coping (Taylor etal., 1983) and may therefore be most importantearly in the adjustment process. As predicted,time since surgery was significantly negativelycorrelated with the downward comparisonvariable (r = -.20, p < .05), such that womencloser in time to surgery are more likely thanothers to compare downward.

Discussion

Our major purpose was to test the predic-tions of four theoretical models concerning thesocial comparisons of victims, and to do so byusing both closed-ended measures of socialcomparison and spontaneous responses. Bothtypes of measures yielded results fruitful tothe hypothesis testing.

The Supercoper Perspective

The supercoper perspective was largely dis-confirmed. The results suggest that direct, per-sonal contact with other breast cancer patientsoccurs more frequently than Taylor and Levin(1976) proposed. However, Taylor and Levindeveloped their ideas in 1974, when discussionof cancer was less open than it is today. Theresults may be an encouraging signal that nowthere is less stigma associated with cancer andthat women now have more opportunities forcontact with other patients. Hence today'scancer patient may also be less susceptible tomedia portrayals of supercopers. In fact, veryfew women even mentioned media figuresduring their interviews, and none of them ap-peared to feel comparatively inadequate. It ispossible that well-known supercopers may beregarded as irrelevant to one's self-evaluation,which is consistent with the idea that potentialcomparison others may be ignored when they


are dissimilar (Goethals & Darley, 1977).Brickman and Bulman (1977) noted that whena superior other is perceived as dissimilar, itcan "take some of the sting out of defeat orinferiority" (p. 162).

Although famous media supercopers maybe ignored, contact with media sources wasnonetheless associated with lower self-evalua-tions of adjustment. This finding has at leasttwo plausible meanings. One is that womenwho are coping poorly seek information fromthe media arid hence have more media contact.The second is that media contact results inlower self-evaluations. If the latter is true, per-haps the media affect self-evaluations notthrough their portrayals of prominent figures,but through their frequent depictions of "ev-eryday" supercopers. Magazine stories aboutbreast cancer often focus on "ordinary"women who are presented as strong and well-adjusted. Although many are described ashaving some anxiety, by the end of the articlethey have usually bounced back, even betteradjusted than before their illness. These su-percoping "ordinary" women may be less eas-ily ignored than famous media figures becausethey are more similar to the patient. Becausethe media seem to have a general tendency topresent victims who are adjusting easily, theseresults may well have implications for threat-ening events other than cancer (cf. Wortman& Dunkel-Schetter, 1979).

The Similarity and Upward ComparisonPerspectives

Festinger (1954) initially proposed that in-dividuals evaluate themselves against similarothers. Our results contradict this hypothesisin some respects and support it in others.When comparing their psychological adjust-ment, respondents rarely compared themselveswith others who were similarly adjusted or whowere similar in related attributes. And theyvirtually never compared themselves withothers who were similar in their physical statusor life situations. At the same time, however,respondents' comparisons were almost alwaysto fellow cancer patients. Although certaincategories (e.g., good coper, others cope worse)permitted comparisons with victims of otherphysical illnesses, by far most comparisonswere made to cancer patients specifically.

About 19% of the respondents did makedownward dimensional comparisons withpeople who did not have cancer, but even theseare similar comparisons by virtue of their beingwith fellow victims of physical illness or dis-ability. (Only two comparisons were not withfellow victims). It seems likely that some rea-sonable range of similarity is required for acomparison to be relevant, but that within thatcontext very similar others may or may not befavored, depending on one's comparison goals.In this study, although most comparisons werewith cancer patients, very few were with verysimilar or with relatively advantaged cancerpatients. In this sense, these results are con-sistent with those of studies that demonstratea decline in similar or upward comparisonswhen there is a threat to self-esteem.

The Downward Comparison Perspective

Overwhelmingly, the comparison of choiceof these breast cancer patients was a downwardcomparison. These findings would not be par-ticularly interesting if the sample were unusu-ally well adjusted, for there would be littlechance of comparing upward. However, datareported earlier suggest that our sample's psy-chological and physical adjustment is typicalof breast cancer patients, and therefore thereshould be available as many potential upwardor similar comparisons as downward compar-isons. That downward comparisons were soprevalent suggests that respondents were veryselective in making comparisons. Downwardcomparisons may help the victim minimizeher sense of victimization (Taylor et al., 1983;Wills, 1981).

Predictors of Downward Comparison

Wills (1981) and Taylor et al. (1983) pro-posed that threat leads to downward compar-isons. However, several analyses were unsup-portive of this position. A possible explanationis that the experience of breast cancer itselfraises threat to a "ceiling." With a restrictionon the variance of threat, it would be impos-sible to detect a covariation with the downwardcomparison index. A related explanation im-plies a ceiling on downward comparison:Downward comparisons may rise rapidly asthreat increases, and may soon reach asymp-tote. Supporting this explanation are the very


restricted ranges for each of the individualdownward comparison variables, which wereessentially dichotomous.

These predictions concerning the role ofthreat are somewhat paradoxical: As one's sit-uation worsens, there may be less opportunityfor downward comparison because the numberof others who are more disadvantaged dwin-dles. This reasoning may explain why thethreat hypothesis was not confirmed. However,contrary to this reasoning, the way in whichrespondents could conceive of a relatively dis-advantaged other was at times quite impressive.For example, one woman who had metastaticdisease considered herself lucky because shehad found peace and meaning in her life, twothings that many other people never achieve.

Lastly, it is possible that threat does not, infact, prompt downward comparison. However,results of laboratory research persuasivelydemonstrate that upward comparisons are thenorm, except when subjects are threatened(e.g., Hakmiller, 1966). Given this evidenceand the plausible explanations listed earlier, itwould be premature to conclude that threat isnot important. Despite the lack of support forthe threat hypothesis, our results strongly sup-port the downward comparison perspective.Downward comparisons were much more fre-quent than other comparisons for each of thecontent domains examined, for each specificvariable in the domains, and for the over-whelming majority of patients. These datafrom breast cancer patients bolster suggestiveevidence of downward comparisons from rapevictims (Burgess & Holmstrom, 1979) anddisabled people (Schulz & Decker, 1985).

One predictor of downward comparison wasidentified: Downward comparisons seemed tobe most important early in the adjustmentprocess. Their function may be to prevent thevictim from being overwhelmed by her new,frightening circumstances (cf. Taylor et al.,1983).

Limitations of the Study

One should keep several considerations inmind when interpreting our results. First, thesample was disproportionately Jewish, and wasmore likely to be of a higher socioeconomicstatus (SES) than is the general population. Itis not clear how these factors would affect the

results, although it is plausible that higher SESindividuals may have more people in thedownward direction with whom to compare.However, another opinion (e.g., T. A. Wills,personal communication, March 18, 1983) isthat economically disadvantaged people are themost likely to make downward comparisons,because they are more threatened.

It is possible that the data are biased by so-cial desirability in that the respondents mayhave made downward comparisons in order to"look good" to the interviewer. However, thispossibility seems less plausible when one con-siders that respondents spontaneously madedownward comparisons in domains other thanthe ones our structured questions involved.Although respondents may have tried to appearwell adjusted by contrasting themselves withpoor copers, it seems unlikely that describingothers who are less fortunate on such dimen-sions as physical status would be seen as so-cially desirable. In addition, people are oftenambivalent about downward comparisons, asWills (1981) and others (Brickman & Bulman,1977) have noted; it is considered less thanadmirable to draw satisfaction from others'misfortune. Moreover, the interviewers weresurprised at the degree to which respondentsrevealed personal and uncomplimentary as-pects of their experience throughout the in-terview. Nonetheless, these considerations re-main important ones, and future researcherswho use other methods may judge the seri-ousness of these problems.

Investigator bias is also potentially trouble-some because we did much of the interviewingand data coding. However, this problem seemsto be minimal. We investigated four theoreticalperspectives that at times generate contradic-tory predictions, and we had ventured no priorhypotheses as to which might be supported.The one perspective with which one of us wasassociated, namely, the supercoper perspective,was largely disconfirmed.

Lastly, because this study is correlationalrather than experimental, directions of cau-sality cannot be definitively established. Forexample, we could not demonstrate that threatcaused downward comparison in this sample.Our primary purpose was to describe the typesof comparison made by these cancer victims,however, and a preponderance of one typeshould support its corresponding theoretical


perspective. The frequency of downward com-parisons in this sample nicely complementsresults of laboratory research, which do con-vincingly show the causal role of threat.

Implications for Social Comparison Theoryand Research

The results of this study have several im-plications for research and theory in socialcomparison. First, this field study suggests thatlaboratory studies should be supplemented bymore naturalistic investigations in order tocapture fully the varied processes and purposesof comparison. Our content analysis of freelyoffered comparisons is a method that is newin the area.9 One may argue that such measuresare only indirect indicators of social compar-isons. However, one also may argue that free-response measures are more direct than themeasures typically used: information seekingand the desire to affiliate. The construct valid-ity of both of these measures has been seriouslyquestioned because both information and af-filiation may be sought for reasons other thansocial comparison (see, e.g., Dakin & Arro-wood, 1981). In contrast, when our respon-dents made a comparative statement, there isno doubt that social comparison was involved.

Our content-analytic method also allows thetexture of social comparison to emerge freely,unconstrained by structured, closed-endedmeasures. The emergent texture expands ourknowledge about the types of comparisonspeople make, and identifies two types of com-parisons not previously detailed in the litera-ture. First, many women compared their ad-justment with a normative standard of ad-justment; they maintained that their ownadjustment was very good compared with thatof fellow sufferers. These comparisons seem tobe made not to specific targets, but to a fab-ricated standard. Second, over 64% of the re-spondents spontaneously made downward di-mensional comparisons, which allow one tochoo.se any dimension one desires when draw-ing a comparison. Unlike the typical down-ward comparison, these two types of compar-ison do not depend on having a disadvantagedcomparison target available, which suggeststhat the literature's emphasis on comparisontargets may be misplaced. In fact, more womenmade downward dimensional comparisons

than any type of comparison involving targets.Target-free comparisons allow the individualmore flexibility in comparison making thanresearchers typically assume.

One may argue that dimensional compari-sons are not necessarily self-selected. The vic-tim's loved ones may point out to her the di-mensions on which she is advantaged in orderto lift her spirits. However, it is very unlikelythat she is not already aware of the possibleupward comparisons (e.g., mastectomees knowabout less disfiguring surgeries). Yet women inour study very rarely focused on these corre-sponding upward comparisons, which suggeststhat they were being selective. Nonetheless, itseems likely that social processes can contrib-ute to the availability of downward compari-sons.

Our findings also emphasize an active, cog-nitive nature to social comparison that mayeven involve construction. Although the lit-erature has emphasized selection of compar-ison others, one's capacity to be selective isgenerally regarded to be constrained by theavailability of comparison targets. In contrast,our findings suggest that availability of com-parison others need not dictate choice of com-parison. First, as described earlier, the respon-dents had many upward and similar compar-isons available, yet they overwhelmingly madedownward comparisons. Second, the respon-dents did not confine their comparisons tospecific targets; at times they seemed to inventcomparison targets, and they were able to ar-rive at comparison dimensions on which theyappeared advantaged, even when other di-mensions on which they were unfortunateloomed quite large. These results suggest thatpeople may, at least under certain circum-stances, manipulate the target or dimension toachieve the outcome they want. At times, then,social comparison may not be particularly"social" at all, in that one's comparisons maynot necessarily involve actual comparisonswith another real human being. Rather, indi-viduals may construct their comparison world.These results portray social comparison asbeing more heavily cognitive than results ofprevious research have suggested.

9 Frey and Ruble (1985) recently used a similar methodin an interesting study of children's spontaneous compar-


Lastly, the respondents' spontaneous andfrequent comparisons suggest that social com-parison is a coping strategy of no small im-portance. Our results contrast with much ofcoping research, which tends to focus on vic-tims' hardships, and suggest that more atten-tion should be paid to victims' active attemptsto rise above adverse circumstances.

References

Brickman, P., & Bulman, R. J. (1977). Pleasure and painin social comparison. In J. M. Suls & R. L. Miller (Eds.),Social comparison processes: Theoretical and empiricalperspectives (pp. 149-186). Washington, DC: Hemi-sphere.

Burgess, A. W., & Holmstrom, L. (1979). Rape: Crisis andrecovery. Bowie, MD: Brady.

Cohen, E, & Lazarus, R. S. (1979). Coping with the stressesof illness. In G. C. Stone, F. Cohen, & N. E. Adler (Eds.),Health psychology—A handbook (pp. 217-254). SanFrancisco: Jossey-Bass.

Cottrell, N. B., & Epley, S. W. (1977). Affiliation, socialcomparison, and socially mediated stress reduction. InJ. M. Suls & R. L. Miller (Eds.), Social comparison pro-cesses: Theoretical and empirical perspectives (pp. 43-68). Washington, DC: Hemisphere.

Dakin, S., & Arrowood, A. J. (1981). The social comparisonof ability. Human Relations, 34, 89-109.

Derogatis, L. R. (1975). The Global Adjustment to IllnessScale (GA1S). Baltimore: Clinical Psychometric Re-search.

Derogatis, L. R., Abeloff, M. D., & Melisaratos, N. (1979).Psychological coping mechanisms and survival time inmetastatic breast cancer. Journal of the American Med-ical Association, 242, 1504-1508.

Dunkel-Schetter, C., & Wortman, C. B. (1982). The inter-personal dynamics of cancer: Problems in social rela-tionships and their impact on the patient. In H. S. Fried-man & M. R. Dimatteo (Eds.), Interpersonal issues inhealth care (pp. 69-100). New York: Academic Press.

Festinger, L. (1954). A theory of social comparison pro-cesses. Human Relations, 7, 117-140.

Frey, K. S., & Ruble, D. N. (1985). What children saywhen the teacher is not around: Conflicting goals in socialcomparison and performance assessment in the class-room. Journal of Personality and Social Psychology, 48,550-562.

Goethals, G. R., & Darley, J. M. (1977). Social comparisontheory: An attributional approach. In J. M. Suls & R. L.Miller (Eds.), Social comparison processes: Theoreticaland empirical perspectives (pp. 259-278). Washington,DC: Hemisphere.

Cruder, C. L. (1977). Choice of comparison persons inevaluating oneself. In J. M. Suls & R. L. Miller (Eds.),

Social comparison processes: Theoretical and empiricalperspectives (pp. 21-42). Washington, DC: Hemisphere.

Hakmiller, K. L. (1966). Threat as a determinant of down-ward comparison. Journal of Experimental Social Psy-chology, Supplement I , 32-39.

Locke, H. J., & Wallace, K. M. (1959). Short marital ad-justment and prediction tests: Their reliability and va-lidity. Marriage and Family Living, 21, 251-255.

Mechanic, D. (1977). Illness behavior, social adaptation,and the management of illness. Journal of Nervous andMental Disease, 165, 79-87.

Pearlin, L. I., & Schooler, C. (1978). The structure of cop-ing. Journal of Health and Social Behavior, 19, 2-21.

Schachter, S. (1959). The psychology of affiliation. Stanford,CA: Stanford University Press.

Schulz, R., & Decker, S. (1985). Long-term adjustment tophysical disability: The role of social support, perceivedcontrol, and self-blame. Journal of Personality and SocialPsychology, 48, 1162-1172.

Silver, R. L., & Wortman, C. B. (1980). Coping with un-desirable life events. In J. Garber & M. E. P. Seligman(Eds.), Human helplessness: Theory and application (pp.279-375). New York: Academic Press.

Singer, J. E. (1966). Social comparison—progress and is-sues. Journal of Experimental Social Psychology, Sup-plement 1, 103-110.

Taylor, S. E. (1983). Adjustment to threatening events: Atheory of cognitive adaptation. American Psychologist,38, 1161-1173.

Taylor, S. E., & Levin, S. (1976). The psychological impactof breast cancer: A review of theory and research. SanFrancisco: West Coast Cancer Foundation.

Taylor, S. E., Lichtman, R. R., & Wood, J. V. (1984). At-tributions, beliefs in control, and adjustment to breastcancer. Journal of Personality and Social Psychology,46, 489-502.

Taylor, S. E., Wood, J. V., & Lichtman, R. R. (1983). Itcould be worse: Selective evaluation as a response tovictimization. Journal of Social Issues, 39, 19-40.

Wheeler, L., & Zuckerman, M. (1977). Commentary. InJ. M. Suls & R. L. Miller (Eds.), Social comparison pro-cesses: Theoretical and empirical perspectives (pp. 335-357). Washington, DC: Hemisphere.

Wills, T. A. (1981). Downward comparison principles insocial psychology. Psychological Bulletin, 90, 245-271.

Wilson, S. R., & Benner, L. A. (1971). The effects of self-esteem and situation upon comparison choices duringability evaluation. Sociometry, 34, 381-397.

Wortman, C. B., & Dunkel-Schetter, C. (1979). Interper-sonal relationships and cancer: A theoretical analysis.Journal of Social Issues, 35, 120-155.

Zanna, M. P., Goethals, G. R., & Hill, J. F. (1975). Eval-uating a sex-related ability: Social comparison withsimilar others and standard setters. Journal of Experi-mental Social Psychology, 11, 86-93.

Received October 29, 1984Revision received June 13, 1985 •

social comparison in adjustment to breast cancer

Documents