Personality and Social Psychology Review

(1)

http://psr.sagepub.com

Personality and Social Psychology Review

DOI: 10.1207/S15327957PSPR0702_102-128 2003; 7; 102 Pers Soc Psychol Rev

Christine R. Harris

Responses, Interpersonal Violence, and Morbid Jealousy

A Review of Sex Differences in Sexual Jealousy, Including Self-Report Data, Psychophysiological

http://psr.sagepub.com/cgi/content/abstract/7/2/102 The online version of this article can be found at:

Published by:

http://www.sagepublications.com On behalf of:

Society for Personality and Social Psychology, Inc.

can be found at:

Personality and Social Psychology Review Additional services and information for

http://psr.sagepub.com/cgi/alerts Email Alerts:

http://psr.sagepub.com/subscriptions Subscriptions:

http://www.sagepub.com/journalsReprints.nav Reprints:

http://www.sagepub.com/journalsPermissions.nav Permissions:

http://psr.sagepub.com/cgi/content/refs/7/2/102 Citations

(2)

Personality and Social PsychologyRevieew 2003,Vol. 7. No.2, 102 128

LawrenceErlbaum Associates, Inc.

A Review of Sex Differences in Sexual Jealousy, Including Self-Report Data, Psychophysiological Responses, Interpersonal Violence, and

Morbid Jealousy

Christine R. Harris Centerftor

Brain & Cognition University ofCalijfrnia, SanDiego

The .specific innate modular theory of jealousy hypothesizes thatnatural selection

shapedsexualjealousy^{as a}mechanismtopreventcuckoldry, and emotional jealousy

as amechanismto preventresourceloss. Therefore,menshould be primarily jealous

overa mate's sexual infidelity and women over a mate's emotional infidelity. Five lines of evidence have been offered as support: self report responses,

psychophysiological data, domestic violence (including spousal abuse and homicide), andmorbidjealousycases.This article reviews each line of evidenceandfind~s onlyonehypotheticalmeasureconsistentwith the hypothesis. This, however, is con-

tradicted bya varietyofothermeasures(including reportedreactionstorealinfidel- ity). Ameta-analvsis ofjealousy-inspiredhomicides, taking into accountbase rates formurder,foundnoevidence that jealousydisproportionatelymotivates^mentokill.

Thefindings^arediscussed froma social-cognitive theoretical perspective.

Several evolutionary psychologists have proposed anintriguing theory about sex differences injealousy over a mate's infidelity (Buss, 2000; Symons, 1979;

Wilson & Daly, 1992). According to these theorists, jealousy in thetwo sexes isafundamentally different phenomenon forreasonsthatcan beunderstoodonlyin light of theprocessof natural selection.Men arehypoth- esized to be innately predisposed to be upset over a mate'ssexualinfidelity, andwomen over a mate's emotional infidelity. This difference, it is argued, reflects different selectivepressures that are assumed tohave operatedonthetwosexes in theancestral environment.

For ourmaleancestors, themajorthreat toDarwinian fitnesswascuckoldry. Ifamate'sinfidelity resultedin pregnancy, a manlosta chance to reproduceandalso risked spending resources on anotherman's progeny.

Male sexualjealousy, accordingto this view,reflects the operation ofapsychologicalmoduleengineeredtomo- tivate behaviors that, in the ancestral environment, wouldhavefunctionedtoprevent cuckoldry. A woman, however,cannotbe tricked intobringingup anoffspring notherown.Hence, itisclaimed, present-daywomen

This work was supported by NSF Grant BCS-9983487 and NIMH GrantROl-MH61626. I amgratefultoall theauthors who sokindlymadeadditionalinformationontheir studiesavailableto me. I also thankWendyWood and LynnMiller forhelpfulcom- ments and acknowledgeNoriko Coburn for her invaluable assis- tance with data coding.

Requestsforreprintsshouldbesent toChristine R.Harris,De- partmentofPsychology-0109, Universityof California SanDiego, LaJolla,CA 92093-0109.E-mail:charris&,)psy.ucsd.edu

should not beparticularly concernedover amate's sexualinfidelity.AsDaly and Wilson (1988)putit

ADarwinianperspectiveonsexual jealousy suggests thehypothesis that it will provetobeasexuallydiffer- entiatedstateinpeople ... becauseof theasymmetri- calrisk of cuckoldry. Whilewomenmaybeexpected tobejealous of their mate's allocation of attention and resources,for example, they donothave the same ra- tionale asmen for being concerned withspecifically sexualfidelity. (p. 182)

For a woman, infidelity posed a different risk to Darwinianfitness:thelossof a mate's resources for her and her offspring. A mate's emotional involvement with another ishypothesizedtohave beenagoodindi- catorof this threat. Therefore, itisargued,present-day womenshouldbeparticularlyupset overamate'semo- tional infidelity. Men, not having faced this adaptive problem, should be lessconcernedaboutemotionalin-

fidelity. Thus both men and women have ajealousy module, but thetwo sexes havedifferenttriggers that activate that module. In sum, "since men and women have evolved different sexual strategies, they should getjealous, angry, and upset about different events"

(Buss,2000,p. 52).

Thisviewofjealousyas asexuallydimorphicadap- tationhasgreatintellectualappeal, seemingly linkingan importantaspectofhumanemotion totheultimateforce shaping living organisms: namely, Darwinian evolution. Notsurprisingly, therefore,the accountiswidely discussedin currenteditions of socialpsychologytext- 102

(3)

SEXDIFFERENCES IN JEAL OUSY9

books and has receivedenormousattentioninbooks and articleswrittenfor thegeneralpublic.Forexample, R.

Wright (I1994), in his critically acclaimed volume The MoralAnimal, citedthejealousy theoryas aparticularly well-worked-out andcompelling example of evolution- arypsychology. Similarly,inhis influential bookHow the Mind Works, Pinker (1997) described sex differences insexualjealousyas anestablished fact. This the- oryalsoserves asthefoundation for Buss's(2000) re- centpopularpressbookonjealousy.

This article offersanalternativeperspective on this body ofwork. The variousdiverse strands of evidence takento support sex differences injealousy are critically examined. Itisargued that although the theory of jealousy proposed by researchers such as Daly and Wilson(1988) deserves admiration for its boldness, the evidence supporting this theory is far less conclusive than isoften maintained. It is suggested that, onbal- ance,the dataforthistheory rangefrominconclusive to downright disconfirmatory. The article concludes with an examination ofpossible ultimate and proxi- matemechanismsthat wouldnotrequire that there bea marked sexdifference injealousy, and with adiscus- sionof the social-cognitive theory of jealousy.

Before turning totheevidence,a comment on ter- minology is necessary. The theory under discussion here is often termed the evolutionary theory of jealousy. A number of contemporary evolutionary psy- chologistsview the humanmindas madeupofmany distinctmodules, each designed by natural selectionto solve aspecific recurring adaptive problemin our ancestral past. Although this might beso, the theory of evolution doesnotentail suchaconclusion. Naturalse- lection might instead have shaped much more do- main-general mechanisms, even mechanisms as general as those proposed by learning theorists.

Alternatively, evolutionmayhaveshaped mechanisms intermediate in specificity or may have shaped less sexually dimorphic mechanisms (Eagly & Wood, 1999; Harris, 2000; Miller & Fishkin, 1997). There- fore, the account ofjealousy under discussion is referred to as thejealousy as a specific innate module (JSIM) hypothesis. Thistermhelpsmakeclear that this debate isnotabout evolution but ratherabout thespeci- ficity of the mechanisms involved injealousy.

Five types ofevidence have been adduced in sup- portof theJSIM model ofsexdifferences injealousy:

self-reportdata,psychophysiological data, murdersta- tistics, spousal abuse and pathologicaljealousy. This articlecriticallyreviews each inturn.

Self-ReportDataPrimarilyFrom College-Age Populations

Afewearly studies hintedat asexdifferenceinjeal- ousy over infidelity. In interviews with 15 couples,

Francis (1977)suggested sexual infidelityis more of- tenassociated withjealousy formenthanforwomen.

Unfortunately,nostatisticsweregiven, and the sample size was small. In a Dutch sample of people whose mates had engagedin infidelity, Buunk (1984) found thatformen, but notforwomen, scores on ajealousy indexwerecorrelatedwith theattribution that the affair was motivated partly by a need for sexual variety.

TeismannandMosher(1978) found that when askedto role-play a jealous situation, more men than women pickedasexual topic. However,thisfindingwould be consistent withmenhavingagreatertendencytofocus onordiscuss sexual matters ingeneral.

The largest set of evidence taken to support the JSIM hypothesis comes from self-report studies that employ forced-choicequestions originally designed by Buss, Larsen, Westen, and Semmelroth (1992). The majority of these studies have recruited college students as participants. Men and women are asked to imagineacommitted sexualrelationship that they have had, presently have,orwould liketohave.Participants arethen askedtochoose which they would findmore upsetting: a matefallinginlovewithsomeoneelseor a matehaving sexual intercourse withsomeone else. A significantsexdifference in which form ofinfidelity is chosen as mostupsetting hasbeenfound inmoststud- ies employing this forced-choice method, with more men than women predicting that sexual infidelity wouldbe worse (seeTable 1).

Meta-AnalysisofForced-Choice Responses

Given the widespread use of the forced-choice method in the literature and the variations emerging from differentstudies,Iconductedameta-analysis on thismeasure tosynthesize these results. The effect size forsexinresponses totheforced-choice questionwas quantified using log-odds ratio (LOR), as recom- mended by most authors (Fleiss, 1994; Haddock, Rindskopf, & Shadish, 1998). With this measure, the odds ofpickingsexfor maleparticipantsaredividedby the oddsofpickingsexfor femaleparticipants, and the natural log of this quantity is usedtorepresent effect size(see last column in Table 1). If therewere no sex difference, one wouldexpect aLOR ofzero. The results of the 32independent sampleswerecombinedus- ing a random-effects model as described by Rosenberg,Adams,andGurevitch (1997),yieldingan estimatedoveralleffect size of 1.00 (95% confidence interval 0.81 < LOR < 1.19), which is typically described as amoderate effect size (Rosenthal, 1996).

IBuss et al. (1992) originally designed two versions ofthe forced-choicequestion.Whenastudy included both questions, results fromQuestionI wereused in themeta-analysesreportedhere.None of thefindingsarechanged by the substitution of results from Ques- tion2(althoughsomeanalysesproducedslightlysmallereffects).

103

(4)

Table 1. StudiesUsedintheMeta-Analysisofthe Forced-ChoiceHvpothetical InfidelitVData, Including SampleCharacteristics andEffectSiz.efortheSex DifferenceinELchSample

Percentage ChoosingSex

Sample Size as Worse

Age Sexual Sex Effect

Country Group Orientation Men Women Questiona Men Women Difference Sizeb Study

Buss et al. (1992) Buss et al. (1999) Buss et al.(1999) Buss etal. ( 1999) Buunk et al. (1996) Buunk ct al. (1996) Buunk et al.(1996) DeSteno & Salovey(1996) DeSteno &Salovey(1996) DeSteno et al. (2002) DeSteno (2001) Dijkstra, et al. (2001) Gaulin et al. (1997) Gearyetal.(2001) Gearyetal.(1995) Geary et al.(1995) Gearyetal.(1995) Geary et al. (1995) Goldenbergetal.(in press) Goldenberg et al. (in press) Harris& Christenfeld

(1996a) Harris(2002) Harris(2002) Harris(2003) Hupka & Bank (1996) Hupka & Bank(1996) Sheets & Wolfe(2001) Sheets &Wolfe(2001) Voracek(2001) Voracek et al.(2001) Wiederman &Allgeier

(1993)

Wiederman& Kendall

(1999)

U.S. S Heterosexual 133 176

115 116 2

Korea S Heterosexual 98 89

97 86

Japan S Heterosexual 213 100

211 99 2

115 109 2

Germany 0 Heterosexual 100 100

100 100 2

Holland S Heterosexual 102 105

102 105 2

U.S. 0 Heterosexual 73 68

U.S. 0 Heterosexual 5726 17,009

Internet

Holland 0 Gay or Lesbian 138 99

138 99 2

U.S. S Heterosexual 141 244 1

China S Heterosexual 54 41

China S Heterosexual 62 75 1

22 29 2

25 28 2

U.S. 0 Heterosexual 47 47 2

U.S. 0 GayorLesbian 49 44 2

138 217 2

U.S. S Heterosexual 84 162 ^I

42 73 2

U.S. 0 Gay orLesbian 55 27 1

55 27 2

Austria 0 Heterosexual 158 177 U

Austria S Heterosexual 134 105 1

134 105 2

Sweden S Heterosexual 173 203 3

Note:S⁼college-agesamples;0=samples thatinclude participants older than 25 yearsofage; U⁼Unknown.

aDifferentversions of theforced-choice infidelityquestion were used in different studies: I=imagining your partnerenjoyingpassionatesexual intercourse vs.imaginingyour partnerformingadeep emotional attachment with that other person. (I * means slightwordingchangefromoriginalver- sion); 2=imaginingyour partnertrying differentsexualpositions vs. imagining your partner falling in love with that other person; 3=othersimilar forced-choice scenario created for thatparticularstudy.

'EfftectSizeexpressedaslog-oddsratio.

104

49 19

76 32

43 11

59 18

53 22

38 13

33 14

60 17

44 12

27 15

30 8

51 3(0

24 14

51 25

58 38

54 34

63 45

32 51

23 26

42 26

73 37

53 23

20 5

54 19

23 9

55 34

41 7

46 22

24 14

47 22

26 11

12 5

61 24

56 18

37 27

48 26

55 30

29 11

24 22

5 4

33 25

26 12

18 7

64 41

62 37

30 44 32 41 31 25 19 43 32 12' 22 21 10 26 20 20 18 19 3 16 36 30 15 35 14 21 34 24 10 25 15 7 37 38 10 22 25 18 2 l 8 14 1 1 23

1.39 1.73 1.62 1.71 1.28 1.19 .94 1.78 1.58 .72 1.42 .85 .60 1.12 .77 .79 .71 -.77 -.17 .72 1.43 1.37 1.30 1.64 1.01 .80 1.97 1.04 .62 1.14 .99 .97 1.59 1.74 .46 1.01 1.02 1.22 .08 .37 .36 .85 .99 .90

25 .99

(5)

SEXDIFFERENCES IN JEALOUSY?

The exclusion ofsamples thatweremadeupofindivid- uals withagayand lesbianorientationraised this value

very slightly to 1.09 (95% confidence interval 0.92 <

LOR< 1.27).

Next,hierarchicalmeta-analyseswereconductedto examinewhether the LOR of the effect ofsex wasmodu- latedby three study variables. The first study variable

wassexualorientation,whichcompared thesexeffectin

thethreegayand lesbianpopulationstotheotherpopu-

lations.2 This revealedasignificant effect of orientation, Q(l)=15.3,p <.0001, reflectingastrongersexeffectfor heterosexualparticipants(1.09) than forgayand lesbian participants (-0.26). The threegayandlesbianpopula-

tionswereexcluded from the remaining analyses. The secondanalysis examinedage, categorizedassamples that primarily included college-age students (sample withameanagelessthan 26) versussamples that included olderindividuals. Therewas asignificant effect of^age with a stronger sex effect for college students (meanLOR=1.20)ascomparedtoolder samples (mean LOR=0.67), Q(1)= 12.3,p<.001. Ideally,wewould have liked to test the effect of country on the forced-choice responses ina meta-analysis. Unfortu- nately, other than theUnited States,mostcountries only had one sample. An analysis using the less optimal grouping by region (United Statesvs.Asiavs.Europe), didnotrevealasignificant modulation of regiononthe

sexeffect,Q(2)⁼3.9,p=.14, although the relativepau-

city of non-U.S. samples leaves that issueopen.Insum, acrosssamples there doesappeartobea sexdifference when the forced-choice hypothetical method is used with heterosexual samples. This effect, however, is greatly reduced in samples thatareolder than thetypical collegeage.Theseresultsarefurtherinterpreted later in thisarticle.

Self-Report Data ThatareInconsistent WithJSIM

Some aspectsof theforced-choice dataare consis- tentwithJSIM. However, thereareseveral findings (or lackthereof) that raise serious doubts about thevalidity of theforced-choice self-report items toassess actual jealousy. In order for the self-report datatooffer sup-

port for JSIM, at least two conditions must be met.

First, there should be convincing evidence that these

measuresdoindeedmeasurejealousy. Second, the evidence shouldfavor the view thatthe differences arise from innate sexually dimorphic specific mechanisms (e.g., rather than noninnate mechanisms ormore gen-

eral innate^sexdifferences). One of thetestsofwhether

ameasureisvalid istofindconverging evidenceacross

many studiesandmultiple operationalizations. As we

shall see,attemptstofindconvergingevidence for the validity of theself-report forced-choicemeasureshave

not supported the JSIM model or the use of the forced-choicemeasure.

First, data are steadily accumulating that suggest that the sex differences predicted by the JSIM model

are rarely found when measures other than the forced-choice items are used to assessjealousy over

hypothetical infidelity. DeSteno and Salovey (1996) foundnoevidence ofasex-by-type-of-infidelityinter-

actionwhenparticipants used continuous rating scales

toestimate theirupsetoverthetwoformsof infidelity.

In another study, the two sexes did notdiffer in their ratingsof theacceptability of differenttypesofinfidel- ity-for example, sex without emotional betrayal and vice versa (Sheppard, Nelson, & Andreoli-,Mathie, 1995). Sheets and Wolfe (2001) hadparticipant's pick which of the twoforms ofinfidelity would be worse,

but then also asked them to rate how much worse.

Analysis of this later question revealed thatonaverage

both men and women rated emotional infidelity as more distressing than sexual infidelity (although

womendid sotoagreaterdegree). DeSteno, Bartlett, Braverman, and Salovey (2002) found no sex differ-

ences on a continuous measure. When participants from^sevennations^wereaskedhowjealous they would feel ^{over a} mate having "satisfying sexual relations with ^someone else," men did nothave higher ratings than women(Buunk & Hupka, 1987). These data are

hardto explain ifmen and womenhave sexually dimorphic jealousy triggers. One might suppose that theseeffectssimply reflectaceilingeffect, withpartic- ipants reporting maximum distress inresponsetoboth forms ofinfidelity. However, this isnotthecase.

ContrarytotheJSIM view, several studieshavere-

ported results that show both sexesaremorebothered by sexual infidelity. For example, Harris (2002b) foundthat^{women as} well asmenestimated that their upsetoveramate's one-night sexual fling duringva-

cation would be greater than theirupset over a com-

parable emotional betrayal. Based on the JSIM per-

spective, this is the very type of situation that men,

but notwomen, should findparticularly upsetting be-

cause it poses grave Darwinian risktomen (potential cuckoldry) but scarcely any Darwinian risk to

women. In another study that used similar scenarios,

menandwomen alsohad greater^meanratingsofup- set over sexual infidelity as compared to emotional infidelity, although statistics for this comparisonwere notprovided (Wiederman & Allgeier, 1993). DeSteno et al. (2002) similarly found thatboth sexes reported greater distress to sexual infidelity. Shackelford, LeBlanc, and Drass (2000) reported that although

men and^women did notdiffer inthe amount ofjeal-

ousy they anticipated having overamate's sexual infidelity,women didreportgreaterangerand hurt than

105

2Someof the studiesspecificallystatedthat gayand lesbianpar- ticipantswereexcluded.However, many studies do notmentionsex- ualorientation;forthisanalysis,weassumedthatthese sampleswere madeupprimarily of heterosexualindividuals.

(6)

HIARRIS

men.3 Another study found that compared to men, women anticipated reacting more violently to a mate's sexual infidelity (Paul & Galloway, 1994).

Oneexception to these findings comesfrom workby Geary, Rumsey, Bow-Thomas, and Hoard (1995).

Chinese men had higher jealousy intensity ratings over sexualjealousy scenarios than Chinese women.

However, Chinese men's ratings were lower than those of American women, whose ratings were slightly higher than theirmale counterparts (although not significantly so). When Geary et al. (2001) used these same measures in another study of American college students, the infidelity typeby sex interaction predicted by JSIM was notfound. Instead, women reported greater hurt and anger over sexual infidelity than did men.4

Second,eventhe forced-choicehypothetical results donotunequivocallysupportJSIM. Although sexdif- ferencesarefound in the studies conducted in countries other than the United States, the majority ofmenin these studies donotchoose sexualinfidelityas moreupsetting than emotional infidelity. Instead, theirresponse pat- terns are often very similar to those of American women.Forexample,aninspection of Table 1 reveals that on one hypothetical forced-choice question, be- tween70% and 80% of Chinese, Austrian, Dutch, and Germanmenreported thattheywould find emotional in- fidelitymoreupsetting than sexualinfidelity.To assess thevariability ofresponses among menina morequanti- tativefashion, wecompared thepropensity of U.S. as comparedtoEuropeanmen topick sexualinfidelityon theforced-choicequestion using the LOR scale. The effect size for this cultural effecton men was0.90 (stan- darderror[SE]=±0.16), which isquite similar inmag- nitude totheeffect ofsexin the overallmeta-analysis reported previously.Thecomparisonof U.S.toChinese menyieldsan evenlarger effect size(LOR=1.73,with SE±0.19).Inotherwords, Europeanmendiffer from U.S.menin theirresponses tothequestionnairetoabout the same extent (and in the same direction) as U.S.

women,andChinesemendiffereven more.Thisfinding seemsquiteproblematictotheJSIM theory. Proponents of JSIM haveargued that the differences in men's and women's responses to the forced-choice questionre- flect the operation of different sexually dimorphicin- natejealousy mechanisms. The fact that one findsal- most identical differences when European men are comparedtoU.S.menispuzzling if one assumes that forced-choicequestionsaretappinginto innate sex-spe- cificjealousy mechanisms. One would wish to have

strongsupporting evidence before taking effect sizes of comparable magnitudeto represent innatefactorsin one caseand cultural factors in another.

A third problem is that the few studies that have compared theforced-choiceresponses to responses on other measures designedtoexamine sexual andemo- tional jealousy have failedtofindconvergentvalidity.

Forced-choice responses were uncorrelated with responses toquestions regardingamate's actual infidelity in two studies (Harris, 2002. 2003) and to psychophysiological responses in a third study, discussed laterinthis article(Harris, 2000).

Finally, and perhaps most troubling, self-report studies that have examined people's reports oftheir actual experiences with infidelity, rather than responses to hypothetical infidelity, have not revealed evidence for sex differences. For example, male and female college students did notdiffer in their assess- mentof howdamaging amate's actual sexual infidelity was to their primary relationship (Hansen, 1987).

Nordid coeds differ in theirratings of how much they focused on the sexual versus emotional aspects ofa mate's affair (Harris, 2003). Moreover, a study with adults who were older than the typical college age foundthat, onaverage, bothmenandwomen, regard- less of sexual orientation, focusedmore on theemo- tional rather than sexualaspectsofamate's actual infidelity (Harris, 2002).5 In asample of sexually open marriages, wives, more thanhusbands, generally had greater negative perceptions of their spouses' affairs and were specifically more bothered by thinking about their mate having sexual intercourse with another person (Buunk, 1981).

In sum, although hypothetical forced-choice self-reportformatssuggestsexdifferences, alternative methods, including the most fundamental data, namely, reports of actual experiences with infidelity,

reveal evidence that runs counter toJSIM. Moreover, even the forced-choice format fails to document any universaltendency formen tofocusonsexual infidel- itymorethanemotionalinfidelity. A variety of factors arelikelytocontributeto sexdifferences observed with the forced-choice hypothetical questions. In one in- triguingrecentstudy, DeStenoet al. (2002)examined theeffect ofcognitive load(retainingastringof7dig- its in short-termmemory) onresponse patternstothe forced-choiceinfidelityquestion. Theyreasoned that if sexdifferences reflect wired-in andsexually dimorphic emotional tendencies as hypothesized by JSIM, then

3Women alsoreportedgreaterreactions toemotional infidelity than men. However, difference scores (degree of emotion in re- sponsetosexualinfidelity-degreeofemotioninresponse to emotionalinfidelity)werevirtuallyidentical for thetwo sexes.

4Thiseffectwasnotsignificantwhenwomentakingbirth control piillswereexcluded from theanalysis.

5An anonymous reviewer suggested that the wording "How muchdidyou focus onthe emotionalaspectsofyourmate'sinfidelity?"isambiguous.However,evenifthiswerethecase,there isno ambiguityin thequestion"Howmuch didyoufocusonthesexual aspects of yourmate's affair?"Yet, therewasno sexdifferenceon thismeasure either, and if there wasany trendat all, it wasfor womentohaveslightlyhigherratingsthanmen.

106

(7)

SEXDIFFERENCESINJEALIOUSY ?

reducing the opportunity for reflective processing or self-presentationstrategies by means ofcognitiveload should increasethe effect, thereby polarizing the two sexes responses. In fact,ithad little effect onmen'sresponses, but it caused women'sresponses to shift to- ward picking sexualinfidelity. Theresultssuggest that inferences or self-presentation strategies may play a particularly strong role in women's responses to the torced-choicequestions.

Some of thevariance between thesexesmight also reflect on men, particularly young men, being more willingtoendorseitems that reflect the importance of sex tothem. This inturn may befueledbytheirargu- ablystronger sex drive(Baumeister, Catanese, &Vohs, 2001), rather than by aninnatetrigger thatis specific onlyto the emotion of jealousy. Responses may also reflect cultural expectations. Still another factor that appears tocontributetosexdifferences inatleastsome samples is that men and women make different as- sumptions about the implications of hypothetical sex- ualand emotionalinfidelity.Forexample, several stud- ieshave found thatpeoplechoose as worsetheform of infidelity that is likelyto imply the co-occurrence of the other form, labeled the double-shot (DeSteno &

Salovey, 1996)ortwo-for-one (Harris & Christenfeld, 1996a, 1996b) hypothesis (see also Dijkstra et al., 2001,fordatafromagayand lesbiansample).6This in- terpretation fits with the fact thatsex, culture, sexual orientation, and age all impact forced-choice responses.Althoughsomeof the factors describedprevi- ouslymayalso contributetoactualjealousy reactions, othersprobably do not(see Harris, 2003, for dataon this). Thishighlightsthe needforfuture workto move beyond the forced-choice and otherhypotheticalmea- suresandto assesscognitions and appraisals thatoccur overrealinfidelity.

Psychophysiological Studies

One study using psychophysiological measures withcollege students has been takentoprovidepartic- ularly compelling support for the JSIM hypothesis (Buss etal., 1992). When askedtoimagine eithersex- ual oremotional infidelityon the partofa mate, men 61tshould be noted that this variable, whenassessedwith different questionsorscales, doesnotalwaysmediatejealousyresponsesto a

significantdegree (Sheets&Wolfe, 2001; Voracek, Stieger,&Gindl, 2001;Wiederman&Kendall, 1999).Buss etal.(1999)exploredthis issue,using questionsthatexplicitlystatethat emotional or sexual infidelity has occurredintheabsenceof the otherform,andstillfounda sexeffect. However, the sexeffectwasfrequently weakerin these cases, suggesting that theimplications of theco-occurrenceofthe two forms ofinfidelity playssomerolebutcannotcompletelyaccountfor thesexdifference.Itshould also be noted that onepotentialconfound with thisapproachis that a womanhavingsexwithoutbeinginloveis likelytoevoke different moraljudgmentsandinferencesthan a man havingsexwithoutbeinginlove, possiblyaffecting choiceresponses.

showed greater heart rate and electrodermal activity (EDA) to imagined sexual infidelity, whereas, as the study is sometimes described, womenshowed the opposite pattern (e.g., see Buss, 2000; Pinker, 1997).

However, a close examination of this study reveals that,forwomen,onlyone outofthree measures (EDA) actually revealed significantlygreaterreactivity inre- sponse toemotional versussexualinfidelity.

Recent attempts toreplicate and extend the Bussetal.

(1992) results seriously question the robustness and meaning oftheirfindings. With additional measures, Harris(2000) reported thatmendo indeed showgreater signs of autonomic arousal whenimagining sexual infidelity relative to emotional infidelity. However, men also showcomparablygreaterreactivitytosexual than toemotionalimagery thatisdevoid ofinfidelity, raising doubts about whether thegreaterreactivity is reallyin- dicativeofgreaterdistress.Moreover, asmentionedpre- viously, the physiological measures during infidelity imagery wereuncorrelated with self-report measures.

Perhapsmoredamagingtothe JSIMtheory is thefact thatHarris (2000), intwoseparatestudies withsignifi- cantlymorestatisticalpowerthanthe originalBuss etal.

(1992) study, foundnoindication thatwomeningeneral showgreaterautonomic arousaltoemotionalinfidelity imagery thantosexualinfidelityimagery. Further, and contrary topredictions byBuss etal.(1992), femaleparticipants who hadexperiencedasexuallycommittedre- lationship showedgreaterreactivitynot toemotionalinfidelity but rathertosexualinfidelity imagery (apattern of arousal that resembled thatof maleparticipants).An- other recentpsychophysiological study also failed to replicate the original Bussetal.findings (Grice & Seely, 2000). Inthisstudy, results from onlyoneof the three physiologicalmeasures werein the directionpredicted by JSIM;menshowedgreaterheartrateincreasesto sex- ualrelativetoemotionalinfidelity imagery, andwomen showedgreaterheartrateincreasestoemotional relative tosexualinfidelity imagery. However, the opposite ef- fectwasfound for EDA, withwomenshowinggreater reactivitytosexualinfidelity imagery, andmen toemotional infidelity imagery. Electromyography (EMG) failedtoshowsexdifferences.7

Inbrief, the results from three articles, which have included a total of five psychophysiological experi- ments,fail toshow clear evidenceformenexperienc- ing greater reactivity to sexual infidelity, and for 7Virtually identical results were obtained whenphysiological scores werecalculatedaschanges fromanemotional imagerybase- line and a sexualimagerybaseline. However,when analyzedinthis manner, the sexeffectonheart rate was stronger. Onemight wonder howthis relatestotheissue of whether thereactivity being measured reflects distress or sexual interest. Griceand Seely (2000) askedpar- ticipants to imagine "twopeople having sexualintercourse"asthe sexualbaseline. Greater baselinereactivity mighthavebeenelicited had participants beenaskedtoimagine "themselveshavingsexwith their mate" as was done inHarris(2000).

107

(8)

HARRIS

women toemotionalinfidelity. Not only is there a lack ofsupport, but also at least three ofthe results from these studies provide evidence incompatible with JSIM(i.e., evidence thatcould be interpreted as indi- catinganemotional jealousy trigger in men and a sexualjealousy triggerin women).

Homicide Statistics

As wehaveseen,evidencefromself-reportsurveys andpsychophysiologicalresponses fail to provide sup- portforsexdifferences in jealousyelicited by infidelity. However, it could be argued that one should not conclude too much either way from self-reports because they are potentially subject to various biases, particularlywhenrespondents have littleornoexperi- encewith actual infidelity. This point is made by the evolutionary psychologists Margo Wilson and Martin Daly (Wilson & Daly, 1992) indescribing the research reviewed in theprevious section:

The bulk of the data are paper-and-pencil responses of captiveundergraduates toquestionnaire items which may ormay nothave anything to do with anything theyhave everexperienced. ... In contrast with the in- conclusive resultsofself-report studies, thereis little ambiguityaboutsexdifferences injealousy whenone looksatsuchreal-world phenomenaashomicide, wife beating,initiationof divorce, and psychiatric cases of

"morbid jealousy."(p.304)

Iwillturn nowto adiscussion of these frequently cited real-worldphenomena.

Daly, Wilson, and Weghorst (1982)wereamong the firsttoclaim that sexual jealousyonthepartofmenisa principalinstigator of violence in all known human soci- eties. This account is echoed by Pinker (1997), who stated, "The largestcauseof spousal abuse and spousal homicide is sexual jealousy, almost always the man's" (p.

489).Itshould be noted that Buss(2000),amajorpropo- nentofJSIM, partedcompanywith these otherJSIMpro- ponents onthisparticular point.Heargued that such ho- micidesare notduetothe sexualjealousy mechanism but rather dueto menhaving evolved "a mate-killing module....whosefunction isnotthreatordeterrence, but rather the literal death ofamate" (p. 122).

Inevaluatingthisevidence,twokinds of issues need tobeconsidered,both ofwhich have been neglected in discussions of thistopic. One issue isepistemological and the otherempirical.

EpistemologicalIssues: Deviancyand Species-Typical Characteristics

Jealousy-inspired murder and assault, along with clinicalstates ofpathological jealousy,certainly seem to reflect intense emotional passions that probably

have something in common with more "ordinary"

formsofjealousy. For that reason,itmayseem reason- able to considerthesephenomenaas auseful source of evidenceabout"garden-variety" jealousy. However,in evaluating Darwinian explanations that postulate innate sexdifferences encodedinthegenomeof the species,thisstrategy appearsquestionable.

Observations about populationextremes are likely tooffera veryunreliableguidetospecies-typicalchar- acteristics. Consider the case ofheight. On average, human females are shorter thanhuman males, witha difference inthe modeand themedian as well as the mean. This shift in thepopulation undoubtedlyreflects natural selection; for example, one possibility is greater intrasexual competition for mates among ancestral males as compared to females (see Wood &

Eagly, 2002, for other possibilities). Observations at oneextreme,however, donotexhibit this species-typicalsexdifference inanyclear-cutfashion,nordothey shed any light on the difference. Although the most common cause of dwarfism, achondroplasia, is as commoninmalesasfemales, some typesof dwarfism areactually more commonamongmales.Further, the ultimatecausesfor thesexratioatthe lower tail of the height distributionare entirely different from the ulti- matecausesof thespecies-typicalsexdifference in the averagevalues(Staheli, 1998).

Itmay sometimes happen, ofcourse, that extrema faithfullyrepresentthe centraltendencyof thepopulation.However, foraDarwinianaccountofallegedspe- cies-typical propertiestofind bettersupportfrom ob- servationsregardingextremathanfrom datainvolving the "normal range" would seem peculiar, to say the least (see Archer, 2000b; Hupka, 1991, for related points). Furthermore, such aview would be inconsis- tent withSymons's proposition that male sexualjealousy is "relatively invariant" (1979, p. 232). Before building atheory on outliers, onewould needa good explanationfor whyan allegedlywired-indisposition would besuppressedordisguisedinmostindividuals, although showing up clearly in disordered or deviant individuals. One should beespeciallywaryof suchar- guments asthey relateto sexdifferences,because it has been observed that males show greater variability in many dimensions than do females (Geary, 1998;

Hedges &Nowell, 1995).

Previous Research

Intheir review ofjealousyandhomicide, Dalyetal.

(1982)discussed several studies thatattemptedto de- termine the motivating factors behindaseries ofmur- der cases. These studies are taken to support the hy- pothesisthat there isa sexdifference incasesofupset over a mate's sexual infidelity, with male but not female sexualjealousyoftenleadingtohomicide. How- 108

(9)

SEXDIFFERENCESINJEALOIUSY?

ever,throughouttheir discussion of the literature, Daly etal. (1982) tendedto blur three distinctcontentions:

(a)thatjealousyisacommonmotive inhomicidecases inmany orallsocieties, (b) that male jealousy leadsto

homicide proportionally ^more often than does female jealousy, and (c) that male jealousy focuseson sexual betrayal and femalejealousy focusesonemotional betrayal. Dalyetal. made apersuasive case for the first contention-jealousydoesappeartobe^acommon pre-

cipitating factor in murderacross manycultures. How-

ever, almost none of the homicide data presented by Dalyet al.provide supportfor thesecondcontention, that sexual jealousy is more frequently a motive for male murderers than it is for female murderers. Even lessevidenceexistsfor the third claim. Asweshallsee,

thedata fall shortin severalrespects.

TheProblem of Base Rates

Theprimaryproblem with the analysesprovidedby Daly et al. (1982) is thatalthoughthey reported that

menkilloutofjealousymoreoftenthanwomen, they failedtoconsidersexdifferences in baseratesformur-

der. Forexample, the study they discussed in greatest detail investigated 690 nonaccidental homicidescom-

mitted inDetroit, Michigan, in 1972. This study was

originally reported by Wilt (1974). Daly et al. coded additional homicides and added thesetotheiranalysis ofthisDetroit sample. Fifty-eight of these cases, ac-

cording to Daly et al., stemmed fromjealousy con-

flicts.8 Of these, 47werecodedas casesprecipitatedby male jealousy and11by femalejealousy. Dalyetal.in-

terpreted these numbersasindicating that sexualjeal-

ousyisastrongermotive formenthanitis forwomen.

However, as is widely known, male perpetrators outnumber femaleperpetratorsforall forms of violent crime andmanynonviolentonesaswell. Forexample, theNationalCrime Victimization Survey, basedon interviews of83,000 U.S. residents, disclosed that, ^ac- cordingtovictimreports, morethan 80% ofcriminal offenderswere men (U.S.BureauofJusticeStatistics, 1992). This imbalance is also confirmed by surveys

asking people about crimes they have perpetrated (Siegel, 1992). Given thedramatically lower tendency of^women tocommit crimesof violence compared to men,itishardtoseehowthefact thatwomencommit- ted about 21% of jealous murders in the Wilt (1974) study canbetakentoindicate thatfemale sexualjeal-

ousy is arelatively weak sourceofangerand aggres-

sion. By analogous reasoning,onemight conclude that becausewomencommitjust8% of therobberies inthe UnitedStates(U.S.Bureau of JusticeStatistics, 1992),

women have relatively weak acquisitive motivations.

Hence,given women'slowerrateof violent crime,the Detroit findings clearlycannotruleoutthe possibility thatinfidelitycausesroughly thesameamountofrage inbothsexes,with mensimply having alower thresh- old for intense violence (perhaps partly because they anticipate beingmoresuccessfulincarryingoutavio- lentact,orperhaps partlyforotherbiologicalorsocial

reasons; seeSiegel, 1992, fordiscussion).

Indeed, certain details of the Wilt (1974) data clearly reinforce thispoint.Only 40 of the 58casesla- beledasjealousy by Dalyetal.(1982) in their Detroit sample involved real orsuspected sexual infidelityor

rivalry due toa "love triangle." In 30 (75%) of these

cases, male jealousy led to the homicide, and in 10 (25%)femalejealousy ledto thehomicide. However, the remaining 18 cases involved situations in which thereis noindication thatathirdpartywasinvolvedor evensuspected (e.g., the homicidewasmotivated simply bya partnertrying toterminate the relationship).

Dalyet al. pointedoutthat the sex difference iseven greaterinthese types ofsituations (17 maleperpetra- tors vs. 1 female perpetrator) andargued that in both types of situationsjealousy arises as a responseto a

threattoreproductivecapacity. However,a moreparsi- moniousexplanation forthese findings is thatregard- less oftheprovocation,men,on average,respondmore

frequently with violence, or their aggression is of a

greaterintensity (support for this latercontentioncan

be found in meta-analysis of aggression by Archer, 2000a). As noted previously, ^across all known situa- tionsandplaces thevastmajority ofcriminalviolence

iscarriedoutbymen.

Insum,simply taking the total number of homicides committed overjealousy and comparing thepercent-

ageof time theperpetratorwas aman versus awoman

providesnoevidence,onewayortheother, for whether

men and women are differentially upset by a mate's sexual infidelity. At aminimum, what is needed is to take into account the differences in total number of crimescommittedby the twosexesand then examine whether the proportion of homicides committed by

men due to sexual jealousy is significantly different from the proportion of homicides committed by

women dueto sexual jealousy. Daly et al. (1982) ac-

knowledged this at one point andadmitted that their

ownDetroit datafailedtoshow suchaproportionaldif- ference. Forexample, men committed 82% of all the homicides and 81% of thejealousy murders, suggest-

ing that in this sample jealousy was not

disproportionally ^a motive for men. Daly et al. also cited three studies that they claimed do provide evi- dencethatmenandwomencommitmurderoutofsex-

ual jealousy at proportionally different rates (Bohannan, 1960; Harlan, 1950; Mowat, 1966).Actual numbersareonlyprovidedfor the Mowatstudy, which examinesjealousy-inspired murderamong insane in- mates (jealousy was the motive in 57 of 473 112.1%I

109 8Dalyetal. (1982)referred to all of these cases as"Sexual Jeal-

ousyConflicts"(p. 14),whereasWilt(1974)categorizedthese cases simplyas"jealousyconflict."