Sexual Arousal and Masculinity-Femininity of Women

19 Pages·2017·0.49 MB·English

JournalofPersonalityandSocialPsychology ©2015AmericanPsychologicalAssociation
2016,Vol.111,No.2,265–283 0022-3514/16/$12.00 http://dx.doi.org/10.1037/pspp0000077
Sexual Arousal and Masculinity-Femininity of Women
Gerulf Rieger Ritch C. Savin-Williams
UniversityofEssex CornellUniversity
Meredith L. Chivers J. Michael Bailey
Queen’sUniversity NorthwesternUniversity
Studieswithvolunteersinsexualarousalexperimentssuggestthatwomenare,onaverage,physiolog-
icallysexuallyarousedtobothmaleandfemalesexualstimuli.Lesbiansaretheexceptionbecausethey
tend to be more aroused to their preferred sex than the other sex, a pattern typically seen in men. A
separateresearchlinesuggeststhatlesbiansare,onaverage,moremasculinethanstraightwomenintheir
nonsexual behaviors and characteristics. Hence, a common influence could affect the expression of
male-typicalsexualandnonsexualtraitsinsomewomen.Byintegratingtheseresearchprograms,we
testedthehypothesisthatmale-typicalsexualarousaloflesbiansrelatestotheirnonsexualmasculinity.
Moreover,themostmasculine-behavinglesbians,inparticular,couldshowthemostmale-typicalsexual
responses. Across combined data, Study 1 examined these patterns in women’s genital arousal and
self-reportsofmasculineandfemininebehaviors.Study2examinedthesepatternswithanothermeasure
of sexual arousal, pupil dilation to sexual stimuli, and with observer-rated masculinity-femininity in
addition to self-reported masculinity-femininity. Although both studies confirmed that lesbians were
more male-typical in their sexual arousal and nonsexual characteristics, on average, there were no
indications that these 2 patterns were in any way connected. Thus, women’s sexual responses and
nonsexualtraitsmightbemasculinizedbyindependentfactors.
Keywords:sexualorientation,sexualarousal,sex-typedbehavior,masculinity-femininity
Studies with volunteers in sexual arousal experiments indicate 2008b;Rieger,Linsenmeier,Gygax,Garcia,&Bailey,2010;Va-
that women’s sexual orientation is weakly reflected in their rela- lentová & Havlícˇek, 2013). The present research attempted to
tive level of physiological sexual arousal to male and female integrate these two established lines of research findings. We
sexualstimuli.Specifically,womenintheseexperimentsshow,on hypothesizedthatmale-typicalsexualarousaloflesbiansislinked
average,substantialsexualarousaltosexualstimulidepictingboth totheirnonsexualmasculinity.Furthermore,themostmasculine-
malesandfemales.Lesbiansconstituteanexceptiontothisgeneral behavinglesbians,inparticular,couldshowthemostmale-typical
findingbecausetheytendtobemorearousedtotheirpreferredsex patternsofsexualarousal.Thetheoreticalassumptionunderlying
(females) than their less preferred sex (males). This pattern is these predictions was that there are common factors that lead to
male-typicalinthesensethatstrongerarousaltothepreferredsex masculinization of both sexual and nonsexual behaviors in some
is more commonly found in men than women (Chivers, Rieger, women.Byusingapooledsetofdatathatyieldedsamplesof115
Latty,&Bailey,2004;Chivers,Seto,&Blanchard,2007;Rieger and345women(dependingontheconductedanalyses)wetested
etal.,2015;Rieger&Savin-Williams,2012a).Aseparatebodyof thesehypotheseswithrespecttowomen’sgenitalarousalandpupil
research indicates that lesbians are, on average, more masculine dilationtosexualstimuli.
than straight women in their nonsexual behaviors, appearances,
andinterests(Johnson,Gill,Reichman,&Tassinary,2007;Lippa, Female Sexual Orientation and Sexual Arousal
Women’s, unlike men’s, sexual attraction patterns may be less
affected by a partner’s sex and more affected by contextual,
cultural, and social factors (Baumeister, 2000; Diamond, 2008;
ThisarticlewaspublishedOnlineFirstOctober26,2015. Rupp & Wallen, 2008; Savin-Williams, 2005). These variables
GerulfRieger,DepartmentofPsychology,UniversityofEssex;RitchC. include pair bonds, attachment history, educational experiences,
Savin-Williams,DepartmentofHumanDevelopment,CornellUniversity; religious beliefs, and acculturation (Peplau, 2001; Peplau, 2003).
Meredith L. Chivers, Department of Psychology, Queen’s University; J.
Because these variables might alter women’s capacity for sexual
MichaelBailey,DepartmentofPsychology,NorthwesternUniversity.
responsemoresothanmen’s,theycouldleadtogreatervariability
This research was supported by the American Institute of Bisexuality
in women’s reported sexual attraction, arousal and orientation
andtheUnitedStatesDepartmentofAgriculture(NYC-321421).Wethank
(Peplau,2003;Wallen,1995).
AlexandraM.FreundandDavidA.Putsforscientificadvice.
Correspondence concerning this article should be addressed to Gerulf In addition to these differences in their reported attraction pat-
Rieger, Department of Psychology, University of Essex, Colchester C04 terns, women and men can differ in their physiological sexual
3SQ,UnitedKingdomE-mail:[email protected] responses. Based on the responses from volunteers in sexual
265
266 RIEGER,SAVIN-WILLIAMS,CHIVERS,ANDBAILEY
arousalresearch,womenare,onaverage,sexuallyarousedtoboth trauma (Slaughter, Brown, Crowley, & Peck, 1997), the female
male and female sexual stimuli, regardless of their sexual orien- response to any sexual stimulus could have evolved in part to
tation.Contrarily,mostmenaresexuallyarousedtoeithermalesor mitigate this risk. For this mechanism, women may have physio-
females, consistent with their sexual orientation. This sex differ- logical sexual responses to a variety of sexual stimuli, including
encewasdescribedwithbothmeasuresofsexualarousalusedin stimuli representing both consensual and forced sexual acts
thepresentresearch:genitalresponse(Bossio,Suschinsky,Puts,& (Suschinsky & Lalumière, 2011), sexual activities of nonhuman
Chivers,2014;Chiversetal.,2004;Chivers,Roy,Grimbos,Can- primates,andmaleandfemalesexualstimuli(Chiversetal.,2004;
tor, & Seto, 2014; Chivers et al., 2007) and pupil dilation while Chiversetal.,2007).Notallavailabledatasupportsuchultimate
viewing sexual stimuli (Rieger et al., 2015; Rieger & Savin- explanationoffemalesexualarousal(Dawson,Sawatsky,&Lalu-
Williams, 2012a). Across these measures, the link of sexual ori- mière,2015;Lalumière&Sawatsky,2015),butregardlessofthe
entationwithphysiologicalresponsestothesamesexorothersex underlying mechanism, women’s unique physiological sexual re-
is considerably weaker in women than in men because women sponsestoeithersexhavebeenrepeatedlyreported(Bossioetal.,
respondmorestronglytobothsexes,.21(cid:2)rs(cid:2).24,(cid:3).03(cid:2)95% 2014;Chivers&Timmers,2012;Suschinsky,Lalumière,&Chiv-
CI(cid:2).43,and,74(cid:2)rs(cid:2).84,.58(cid:2)95%CI(cid:2).95,respectively ers,2009).
(Riegeretal.,2015). However, women’s sexual responses are moderated by their
“Female-typical” physiological sexual arousal could therefore sexualorientation.Onaverage,straightwomenaremorelikelyto
bedescribedassignificantandmostlynonspecificsexualarousal show no significant differences in their sexual responses to both
to both males and females, regardless of preference, whereas male and female sexual stimuli. In contrast, lesbians are more
“male-typical” sexual arousal as stronger sexual responses to the sexually aroused to same-sex stimuli (women) than to other-sex
preferredsexthantothelesspreferredsex(Chiversetal.,2007). stimuli(men).Thisdifferencebetweenstraightwomenandlesbi-
This difference is not absolute. For example, a proportion of ans is not strong, but has been observed both for their genital
bisexual-identified men are sexually aroused to both males and arousal (Chivers et al., 2004; Chivers et al., 2007; Rieger et al.,
females(Riegeretal.,2013)andareinthissense“female-typical.” 2015) and pupil dilation to sexual stimuli (Rieger et al., 2015;
Althoughthesemenarelikelylesscommonthanmenwithsexual Rieger & Savin-Williams, 2012a). When bisexual women were
orientations and arousal toward one preferred sex (Rosenthal, studied,theywerein-betweenstraightwomenandlesbiansintheir
Sylva, Safron, & Bailey, 2011), these findings highlight that arousalpatternstothesamesexorothersex(Riegeretal.,2015;
“male-typical” and “female-typical” arousal does not apply to all Rieger&Savin-Williams,2012a;Timmers,Bouchard,&Chivers,
menandwomen.Likewise,thereisconsiderablevariabilityacross inpress).Acrossstudiesandmeasures,theassociationofwomen’s
sex in physiological sexual arousal to male or female stimuli sexualorientationwiththeirsexualresponsetothesamesexover
(Rieger et al., 2015; Figures 1–4). Some men and women have theothersexissmallbutconsistent,.21(cid:2)rs(cid:2).24,(cid:3).03(cid:2)95%
sexual responses that are contrary to the general trend. Our de- CI(cid:2).44(Riegeretal.,2015),eventhoughtheeffectcanbemore
scriptionsofsexdifferencesinsexualarousalthereforeapplyonly pronounced if sexual response is assessed with pupil data rather
onaverage,andpartoftheobservedvariabilitycouldbeexplained thangenitalarousal(Rieger&Savin-Williams,2012a).
by other factors than sex, such as the behavioral masculine and One study did not report that lesbians had stronger genital
femininetraitsexaminedinthisresearch. responses to the same sex than other sex (Peterson, Janssen, &
Other measures indicate similar sex differences in sexual re- Laan, 2010), but because this study did not include distinct male
sponse.Theassessmentsviareactiontime(RT;Wright&Adams, andfemalestimuli,itisdifficulttocomparetotheaforementioned
1994; Wright & Adams, 1999), viewing time (Ebsworth & Lalu- studies.Basedonthataforementionedresearch,theoverallfinding
mière,2012;Lippa,2012;Lippa,2013),thermography(Huberman is that lesbians respond physiologically stronger to the same sex
&Chivers,2015),orneuroimagingwhileviewingstimuli(Costa, thantotheothersex.Inasimilarfashion,othermeasuresofsexual
Braun,&Birbaumer,2003;Sylvaetal.,2013)suggestthatwom- response, RT (Wright & Adams, 1994; Wright & Adams, 1999),
en’s responses to sexual stimuli are less linked to their sexual and viewing time (Ebsworth & Lalumière, 2012; Lippa, 2012;
orientationthanaremen’s.Acrossthesemeasures,women,more Lippa, 2013), indicate that lesbians have, on average, greater
than men, respond more strongly to males and females, and less responses to the same sex than the other sex, whereas straight
specificallytotheirpreferredsex. women do not show a difference in their responses. Stronger
Sex-specific selection pressures might explain this general sex sexual arousal to one sex, congruent with someone’s reported
difference in the association of sexual orientation with sexual sexualorientation,isusuallyfoundinmen.Inthissense,lesbians
response. Men have likely evolved with a strong sex drive show a more male-typical sexual arousal pattern compared with
(Baumeister, 2000) and strong sexual arousal toward sexually otherwomen.
relevant targets (Bailey, 2009), and their combination facilitates
prompt sexual responses required for reproduction. Women may
Female Sexual Orientation and Masculinity-Femininity
have evolved to be sexually responsive in sexual context-
dependent situations in order to avoid genital injury. Support for Just as some patterns of sexual arousal are more male-typical
this hypothesis is derived from comparisons across species and andothermorefemale-typical,sodononsexualbehaviorsvaryin
cultures. Forced copulation in several species (Galdikas, 1985; their sex-typicality. Studies on this topic usually fall under the
McKinney,Derrickson,&Mineau,1983;Thornhill,1980)andin rubric of research on “masculinity” and “femininity.” Conceptu-
mosthumansocieties(Palmer,1989;Sanday,1981)indicatethatit alizationsofmasculinityandfemininityhavebeenheavilydebated
may have occurred throughout human evolution (Thornhill & overthedecades(Constantinople,1973;Lippa,2005a;Spence&
Thornhill, 1983). Because forced copulation can lead to genital Buckner,1995).Oneapproachistodefinemasculinityandfemi-
SEXUALAROUSALANDMASCULINITY-FEMININITY 267
ninity as opposite poles of an encompassing psychological and early development (Rahman, 2005; Savic, Garcia-Falgueras, &
behavioraltrait(Lippa,1991;Lippa,2005a;Lippa,2005b;Lippa, Swaab,2010).
2008b). One-dimensional self-ratings of adulthood masculinity- In sum, there is a robust link between sexual orientation and
femininity exhibit correlates pointing to that trait’s construct va- masculinity-femininity, even though the magnitude of the effect
lidity, including correlates with gender-typed occupational and varies by measure (Lippa, 2008b; Rieger et al., 2010; Rieger &
recreational interests (Lippa, 1991; Lippa, 1995a; Lippa, 1995b), Savin-Williams, 2012b). For example, sexual orientation is more
recalledchildhoodmasculinity-femininity(Bailey,Dunne,&Mar- closely linked to self-reports of masculinity-femininity in child-
tin, 2000; Lippa, 2008a; Rieger & Savin-Williams, 2012b), and hoodthaninadulthood;yet,withobserver-ratingsfromthesetime
observer-ratingsofmasculinity-femininityinchildhoodandadult- periodstheoppositetendstothecase(Baileyetal.,2000;Rieger
hood(Lippa,1998;Rieger,Linsenmeier,Gygax,&Bailey,2008; et al., 2008; Rieger et al., 2010). The exact meaning of these
Riegeretal.,2010). differences is unclear. However, because of these variations, any
Thereisapossiblecoretomasculinity-femininitythatcontains relationship of sexual orientation with both sexual arousal and
sexualorientationinadditiontogender-typedself-concepts,inter- masculinity-femininitymightfurtherdependonwhichmeasureof
ests, appearances, vocal patterns, and nonverbal displays (Lippa, masculinity-femininity is used. We examined this possibility in
2005b;Riegeretal.,2010).Thatis,sexualorientationdifferences Study2.
in masculinity-femininity within each sex reflect those usually In addition to variation across measures, there is further varia-
seen between the sexes. In one meta-analysis, lesbians reported tion within measures, which can differ by sexual orientation
moremasculineandlessfeminineinterestsandself-conceptsthan (Lippa, 2005b; Lippa, 2008b; Lippa, 2015; Rieger et al., 2008;
straightwomen;conversely,gaymenweremorefeminineandless Rieger et al., 2010). In a meta-analysis, lesbians were more vari-
masculine than straight men (Lippa, 2005b). These effects were able than straight women in their self-reported masculinity-
large in women and men, 1.28 (cid:2) ds (cid:2) 1.46, 1.18 (cid:2) 95% CI (cid:2) femininityandinsex-typedoccupationalandrecreationalinterests
1.56,and0.60(cid:2)ds(cid:2)1.28,0.50(cid:2)95%CI(cid:2)1.38,respectively. (Lippa,2005b).Differencesinvariationarenotalwaysfound.In
In another meta-analysis, lesbians recalled more masculine and other data from 1,383 women, lesbians were more variable than
less feminine childhood behaviors than straight women; the con- straight women in their self-reported masculinity-femininity, but
versewasfoundforgaymenandstraightmen(Bailey&Zucker, notintheirsex-typedoccupationalinterests(Lippa,2015).Inother
1995).Theseeffectswerealsolarge,d(cid:4)0.96,0.26(cid:2)95%CI(cid:2) studies, lesbians were more variable in their observer-rated
1.66,andd(cid:4)1.31,0.45(cid:2)95%CI(cid:2)3.08,respectively. masculinity-femininity, but not their self-reported adulthood or
Prospective studies suggest that this difference in masculinity- childhood masculinity-femininity (Rieger et al., 2008; Rieger et
femininity can be observed in young children prior to the devel- al., 2010). Thus, although not always confirmed, lesbians can be
opment of their adult sexual orientation (Drummond, Bradley, morevariableintheirmasculinity-femininitythanstraightwomen.
Peterson-Badali,&Zucker,2008;Green,1987;Riegeretal.,2008; When bisexual women were studied, they were intermediate
Steensma,vanderEnde,Verhulst,&Cohen-Kettenis,2013).Sim- between straight women and lesbians in their masculinity-
ilarly, sexual orientation differences in masculinity-femininity in femininity, with lesbians being consistently more masculine and
adulthood can be observed by others based on motor behaviors, lessfemininethanstraightwomen(Lippa,2005b;Lippa,2008b).
speech patterns, and physical appearance (Johnson et al., 2007; This finding corresponds with their aforementioned pattern of
Rieger et al., 2010; Valentová & Havlícˇek, 2013). In addition, physiological sexual arousal, as bisexual women are somewhat
facialfeaturesofstraightandgaymenandwomenaredifferently moremale-typicalintheirarousalthanstraightwomen,butlessso
perceived (Rule, Ambady, Adams, & Macrae, 2008; Rule, Am- than lesbians (Rieger et al., 2015; Rieger & Savin-Williams,
bady, & Hallett, 2009) and it is possible that this difference is 2012a;Timmersetal.,inpress).Thus,apredictionisthatforboth
relatedtomasculinity-femininity. sexualandnonsexualbehaviors,bisexualwomenaremoremale-
Straightandgaymenandwomenfurtherdifferinsomesexually typicalthanstraightwomenbutlesssothanlesbians.
dimorphicneuroanatomicalstructuresandtheirfunctions(Rahman
& Yusuf, 2015). Lesbians and gay men are, on average, more
Female Sexual Orientation, Sexual Arousal, and
similartotheothersexintheirhypothalamicactivationinresponse
Masculinity-Femininity
to human pheromones (Berglund, Lindström, & Savic, 2006;
Savic,Berglund,&Lindström,2005),andintheircerebralasym- The review this far suggests that lesbians are in general more
metry and functional connections (Savic & Lindström, 2008), male-typical than straight women in their physiological sexual
which possibly affect differences in linguistic processing (Rah- arousal (Rieger et al., 2015) and their behavioral masculinity-
man,Cockburn,&Govier,2008).Furthermore,gaymenaremore femininity(Lippa,2008b).Ifthereisacommonfactorthatinflu-
female-typical than straight men in spatial processing such as encesmale-typicalsexualandnonsexualbehaviorsinwomen,then
mental rotation, whereas lesbians are, to a smaller degree, more a hypothesis is that because lesbians are more masculine, on
male-typical than straight women (Rahman & Wilson, 2003b). average, they also show male-typical sexual arousal, on average.
Similar sex-dimorphic differences between lesbians and straight Hence, overall differences in masculinity-femininity between
womenhavebeenobservedfortheirotoacousticemissions,minute women might explain the effect of sexual orientation on female
soundsemittedbytheinnerearthatareusuallymorecommonin sexualarousal.Suchhypothesissuggeststhatwomen’smasculinity-
men than women (McFadden & Champlin, 2000). Neurological femininity mediates the relationship of their sexual orientation with
structures and cognitive functioning may therefore be partly sex- theirsexualarousaltothesameorothersex.
atypicalinwomenandmenwithsame-sexsexualorientations,and Alternatively, an interaction of sexual orientation with behav-
this is possibly due to differentiations of neural circuits during ioralmasculinity-femininitycouldexplainwhysomewomenshow
268 RIEGER,SAVIN-WILLIAMS,CHIVERS,ANDBAILEY
male-typical sexual arousal. As reviewed above, the effect of orientation with masculine behaviors and male-typical sexual
lesbians’ stronger arousal to their preferred sex is small in mag- arousalinwomen.
nitude, and there is considerable variability in women’s arousal Notably,itislittleunderstoodtowhatdegreegonadalinfluences
patterns (Chivers et al., 2007; Rieger et al., 2015). It is therefore affect physiological sexual arousal. Elevated androgen levels in
possiblethatonlysome(butnotall)lesbiansdrivetheeffectthat adulthood can enhance sexual motivation in both males and fe-
linkstheirsexualorientationtostrongersexualarousaltowardthe males (Bancroft, 2005; Jones, Ismail, King, & Pfaus, 2012), but
same sex. There is also considerable variation in behavioral whether they influence male-typical physiological sexual re-
masculinity-femininity that is sometimes (although not always) sponsesineithersexisunknown.However,ifdatasuggestedthat
stronger in lesbians (Lippa, 2005b; Lippa, 2008b; Lippa, 2015; masculine behaviors and male-typical arousal of lesbians are in-
Riegeretal.,2008;Riegeretal.,2010).Hence,somelesbiansare terrelated, such findings would at least be consistent with the
especially masculine, compared both with straight women and proposal that an underlying factor (hormonal or otherwise) ac-
other lesbians. Perhaps these are the women, in particular, who countsforsuchpattern.
respond sexually more to their preferred sex than the other sex.
Thus,themostmasculine-behavinglesbians(comparedbothwith Overview of Studies
straightwomenandlessmasculine-behavinglesbians)couldshow
Basedonthereviewedliterature,thefollowinghypotheseswere
themostmale-typicalsexualarousal.
tested:
Because straight women’s sexual arousal is, in general, not
specificallydirectedtowardmalesorfemales,whereasforlesbians Hypothesis1:Lesbianswillshow,onaverage,strongersexual
thereisatrendformorearousaltowardtheirpreferredsex,wehad arousaltothesamesexthantotheothersex,whereasstraight
lessclearpredictionsforstraightwomenthanforlesbiansregard- womenwillnot,onaverage,differintheirarousaltothesame
inghowtheirmasculinity-femininitycoulddistinguishtheirsexual sexorothersex.
arousal.Thus,thehypothesisaboutaninteractionofsexualorien-
Hypothesis 2: Lesbians will report and show, on average,
tation with masculinity-femininity focuses on the prediction that
greatermasculinityandlessfemininitythanstraightwomenin
for lesbians masculinity-femininity differentiates their sexual
theirnonsexualself-conceptsandbehaviors.
arousalpatterns,whereasforstraightwomenwemadenopredic-
tions. We note, however, that the moderation analyses reported Hypothesis 3: If lesbians are more male-typical than straight
belowincludedtestingforthepossibilitythatstraightwomendiffered women in both their sexual arousal and their nonsexual be-
in their sexual arousal, depending on their masculinity-femininity. haviors,thentherelationshipoffemalesexualorientationwith
These analyses also allowed exploring how masculinity-femininity sexualarousalwillbemediatedbytheirnonsexualmasculine
affectedsexualarousal,regardlessofwomen’ssexualorientations. behaviors.
What factors could explain that lesbians are on average, if not
someoftheminparticular,moremale-typicalinboththeirsexual Hypothesis4:Alternatively,themostmasculine-behavingles-
arousalandnonsexualbehaviors?Bothprenatalandearlypostna- bians, in particular, will show stronger sexual arousal to the
talandrogenexposurepredictmasculinizedbehaviorsintheearly same sex than to the other sex, both compared with straight
development of boys and girls (Auyeung et al., 2009; Lammin- womenandlessmasculine-behavinglesbians.Thus,therela-
mäki et al., 2012). In addition to their effects on sex-typed mor- tionofsexualorientationandsexualarousalwillbemoderated
bynonsexualmasculinebehaviors.
phology(Arnold,2009),theseearlyandrogenexposuresinfluence
masculine behaviors, interests, and cognitive abilities throughout
Thepresentresearchinvestigatedthesehypothesesbycombin-
the life course (Berenbaum & Beltz, 2011). Early gonadal influ-
ingpublisheddata(Chiversetal.,2004;Riegeretal.,2015;Rieger
ences are also prominent candidates for the codevelopment of
& Savin-Williams, 2012a) and unpublished data on the relation-
sexual orientation with masculinity-femininity (Hines, 2011) and
ship of sexual orientation and physiological sexual response. By
for the variation of masculinity-femininity within sexual orienta- merging these data, analyses offered better information on the
tions (Bailey & Zucker, 1995). In one study, lesbians with mas- magnitudeoftheeffectoffemalesexualorientationonphysiolog-
culineself-conceptshadmoremasculineanatomicalfeatures(i.e., icalsexualarousal.Moreover,theeffectofmasculinity-femininity
higher waist-to-hip ratios) than other women, possibly because onsexualarousalhasnotbeenpreviouslyreported.Presentanal-
these women have been exposed to higher levels of androgens yses investigated whether women’s levels of masculinity-
during development (Singh, Vidaurri, Zambarano, & Dabbs, femininitymediatedormoderatedtherelationshipoftheirsexual
1999). Furthermore, these women exhibited higher levels of sali- orientationwithsexualresponse.
vary testosterone; this could also suggest greater developmental Study 1 examined these hypotheses in 115 to 152 women
androgenization, at least to the extent that developmental andro- (numbers varied by analyses) whose genital arousal and self-
gens may be reflected in their levels in later life (Auyeung, reportedadulthoodmasculinity-femininitywereassessed.Study2
Lombardo, & Baron-Cohen, 2013; Romeo, Richardson, & Sisk, tested these hypotheses in 186 to 345 women whose sexual re-
2002;Schulz,Molenda-Figueira,&Sisk,2009). sponse was assessed via their pupil dilation, and for who, in
These gonadal influences, in combination with genetic influ- addition to their reported adulthood masculinity-femininity, self-
ences (Bailey et al., 2000; Burri, Cherkas, Spector, & Rahman, reports from childhood and observer-ratings of adulthood behav-
2011)orevenepigeneticinfluences(Ngun&Vilain,2014)could iorswereavailable.Studies1and2werekeptseparatebecausethe
explainthecodevelopmentofsexualorientationandmasculinity- majority of genital arousal data could only be linked to reported
femininity. They may also account for associations of sexual adulthood masculinity-femininity, and the majority of reported
SEXUALAROUSALANDMASCULINITY-FEMININITY 269
childhood masculinity-femininity and observer-rated adulthood The proportion of being Caucasian did not significantly differ
masculinity-femininitywerelinkedtopupildata.Ifthesedifferent across sexual orientation groups, (cid:5)2(6) (cid:4) 6.73, p (cid:4) .35. As we
measuresyieldedsimilarfindings,itwouldsupporttherobustness reportbelow,thesedifferenceshadlittleeffectontheassociations
of the effect of gender-typed behaviors on the relationship of ofsexualorientation,masculinity-femininity,andsexualarousal.
sexualorientationwithfemalesexualarousal. Measures.
Sexualorientation. UsingKinsey-typescales(Kinsey,Pome-
Study 1 roy, Martin, & Gebhard, 1953), participants either reported their
sexualorientationidentitiesandfantasiesforthelastyearandtheir
Study1combinedtwodatasets(Chiversetal.,2004;Riegeret
adulthood in general (Chivers et al., 2004), or their sexual orien-
al., 2015) to examine patterns of female genital arousal. As de-
tationidentities,attractions,andfantasiesinadulthood(Riegeret
scribed below, these datasets differed in some aspects of their al.,2015).Measureswerehighlycorrelatedineachdataset(ps(cid:2)
methodologies. However, these differences did not statistically .0001,.80(cid:2)rs(cid:2).96,.70(cid:2)95%CI(cid:2).99),andaveragedwithin
moderate the relationships of sexual orientation with genital
participants. For this composite, a score of 0 represented exclu-
arousal or reported masculinity-femininity. For the sake of sim-
sivelystraight,ascoreof3bisexualwithequalpreferences,anda
plicity, and because the overarching goal of this research was to
scoreof6exclusivelylesbian.
examine patterns across all available data, these nonsignificant
Note that even though we focus in our interpretations on a
differences between studies are not reported in the following
difference between straight women (Kinsey scores 0–1) and les-
results.
bians (Kinsey scores 5–6), the data included bisexual women
(Kinseyscores2–4).Inthemajorityofanalyses,bisexualwomen
Method
fell in-between straight women and lesbians in their arousal and
Participants. Advertisements for the studies were placed in masculinity-femininityscores.WeaddressthisfindingintheGen-
newspapersandwebsites,eitherinChicago(Chiversetal.,2004) eralDiscussion.
orclosetoaNortheastuniversity(Riegeretal.,2015).Atotalof Masculinity-femininity. In one dataset (Chivers et al.,
173womenwererecruited.For21ofthesewomen,nosubstantial 2004), women were asked how masculine and how feminine
genital responses were detected (i.e., at least 0.5 SD maximum theywere.Similarbriefquestionsaboutmasculinity-femininity
arousaltoasexualstimulusascomparedwithaneutralstimulus, have demonstrated congruent validity because of correlates
Chivers et al., 2004) and were thus excluded from analyses. with gender-typed recreational and occupational interests,
Excludingtheseparticipantsdidnotaffectthedirectionorsignif- observer-ratings of masculinity-femininity, and reported child-
icanceofresults;ingeneral,exclusionwasstatisticallybeneficial hoodmasculinity-femininity(Lippa,1991;Lippa,1998;Lippa,
because it yielded slightly stronger effects. The remaining 152 2008a). Answers to these questions were given on 7-point
women self-identified as “straight” (n (cid:4) 31), “mostly straight” scales,rangingfrom1(stronglydisagree)to7(stronglyagree).
(n (cid:4) 41), “bisexual leaning straight” (n (cid:4) 14), “bisexual” (n (cid:4) Masculine and reversed feminine scores were correlated, p (cid:2)
10), “bisexual leaning lesbian” (n (cid:4) 18), “mostly lesbian” (n (cid:4) .0001, r (cid:4) .64, 95% CI [.42, .79] and reliable (Cronbach’s
19), and “lesbian” (n (cid:4) 19). The average age (SD) was 24.41 alpha (cid:4) .78). These scores were averaged such that higher
(5.17) years. The most common ethnicity was Caucasian (63%), scores indicated more masculinity and less femininity.
followed by Hispanic (13%), Black (10%), and mixed ethnicities Inthisdataset(Chiversetal.,2004),masculinity-femininitydata
(9%). Education was coded as 1 (no high school), 2 (some high wereavailablefor48outofthe85women.Thisinformationwas
school), 3 (high school diploma), 4 (some college), 5 (college originally collected via a paper questionnaire, and electronically
graduate), and 6 (postgraduate student or degree). The average enteredlongafteralldatacollectionhadbeencompleted;atwhich
education (SD) was 4.46 (.86); the median was 4.00. The most point questionnaire data from 37 women were no longer retriev-
commoneducationwas“somecollege”with62%,with36%being able. Because of this limitation, we report analyses below that
currentlyincollege. investigated possible systematic differences between women for
Distributions of age, ethnicity, and education by sexual orien- whomsuchdatawereavailableormissing.
tation are shown in Table 1. Analyses of variance indicated that In the other dataset (Rieger et al., 2015), adulthood masculinity-
participants with lesbian identities were older than other women, femininitywasreportedbyall67womenwiththe10-itemContinu-
p(cid:4).001,R2(cid:4).10,andhadmoreeducation,p(cid:4).004,R2(cid:4).12. ousGenderIdentityScale,whichexhibitsmodesttostrongcorrelates
Table1
DistributionofSexualOrientationIdentitiesAcrossAges,Ethnicities,andEducationinStudy1
Bisexualleaning Bisexualleaning
Women(N(cid:4)152) Straight Mostlystraight straight Bisexual lesbian Mostlylesbian Lesbian
Number 31 41 14 10 18 19 19
Averageage 22.81[21.18,24.44] 23.08[21.49,24.67] 24.20[21.78,26.62] 22.89[19.60,26.18] 26.24[23.28,29.19] 26.78[23.76,29.81] 27.21[23.92,30.51]
PercentageCaucasian 67[50,80] 63[47,77] 40[20,64] 67[35,88] 52[31,74] 78[58,90] 64[39,84]
Averageeducation1 4.19[4.02,4.37] 4.35[4.08,4.63] 4.60[4.19,5.01] 4.11[2.99,5.23] 4.71[4.27,5.14] 5.00[4.69,5.40] 4.29[3.71,4.86]
Note. Numbersinbracketsrepresent95%confidenceintervals.
1Educationwascodedas1(nohighschool),2(somehighschool),3(highschooldiploma),4(somecollege),5(collegegraduate),6(postgraduatestudent
ordegree).Oneparticipantindicated“othereducation.”
270 RIEGER,SAVIN-WILLIAMS,CHIVERS,ANDBAILEY
withothermeasuresofreportedandobservedmasculinity-femininity Genitalarousaldatawereaveragedwithinstimuliandforeach
(Riegeretal.,2008;Riegeretal.,2010).Statementsincluded:“Most participant and, based on previous recommendations, z-scored
people see me as more masculine than other women” and “My within participants (Harris, Rice, Quinsey, Chaplin, & Earls,
mannerismsarelessfemininethanthoseofotherwomen.”Answers 1992). In each dataset, such standardization within participants
weregivenon7-pointscales,rangingfrom1(stronglydisagree)to7 was conducted across responses to all presented stimuli. In one
(strongly agree). Item-reliability (Cronbach’s alpha) was .91. Re- dataset (Chivers et al., 2004), participants’ average responses to
sponsestoitemswereaveragedandhigherscoresrepresentedgreater the second neutral stimulus (which they viewed after return to
masculinity. baseline) were subtracted from their average responses to sexual
Stimuli. In one dataset (Chivers et al., 2004) sexual stimuli stimuli. In the other dataset (Rieger et al., 2015), average genital
included two male and two female stimuli, showing videos of responsetothe10sprecedingasexualstimulus(i.e.,attheendof
either two males or two females engaged in sexual activities. In aneutralstimulusandatwhichtimetheyhadreturnedtobaseline)
addition,twonaturedocumentarieswereshownfortheassessment wassubtractedfromtheaverageresponsetothisstimulus.Wethen
ofparticipants’baselinearousal.Intheotherdataset(Riegeretal., computed,foreachparticipant,twoaveragevaluesreflectinggen-
2015),3-minvideosofthreemalestimuliandthreefemalestimuli italresponsetosame-sexstimuliandother-sexstimuli.
were used. These videos depicted either a male or female model
masturbating. Six 2-min videos were taken from a nature docu-
mentaryforassessingbaselinegenitalresponses. Results and Discussion
Genitalarousal. ABIOPACMP100dataacquisitionunitand
Hypotheses 1. We predicted that lesbians would be more
the program AcqKnowledge recorded genital responses every 5
genitally aroused than straight women to the same sex than the
ms. Women’s genital arousal was assessed via change in vaginal
othersex.Wefirstinvestigatedarousalpatternsacrossallwomen.
pulse amplitude (VPA) using vaginal photoplethysmographs
One-samplettestsindicatedthatwomenofallsexualorientations
(Janssen,Prause,&Geer,2007).TheVPAsignalwassampledat
respondedonaveragemoretosame-sexstimuli,ascomparedwith
200 Hz and high-pass filtered at 0.5 Hz with 16 bits resolution.
neutral (a score of 0), p (cid:2) .0001, d (cid:4) 2.00 [1.89, 2.12], and to
VPAwasmeasuredaspeak-to-troughamplitudeforeachvaginal
other-sexstimuli,ascomparedwithneutral,p(cid:2).0001,d(cid:4)2.03
pulse. VPA signals indicate changes of vaginal blood flow and
[1.93,2.15].
exhibitbothconvergentanddiscriminantvalidityoffemalesexual
response(Suschinskyetal.,2009). Wethenregressedwomen’sgenitalarousaltothesamesexand
Procedure. Participants provided written informed consent to the other sex onto their sexual orientation. Because it was
andwereseatedinaroomfacingascreen.Inprivatetheyinserted possiblethatresponsestothesamesexthantotheothersexwere
the photoplethysmograph. First, participants watched an adapta- particularly pronounced in exclusively lesbians (that is, for those
tionstimulus(anaturevideo)toestablishbaselineresponse.Next, withKinseyscoresof6),weincludedatestforacurvilineareffect
in one dataset they randomly watched the sexual videos and the in these analyses (Rieger & Savin-Williams, 2012a). Unless oth-
othernaturevideo;betweenvideos,participantsworkedonques- erwise noted, however, these curvilinear effects were not signifi-
tionnairesandmentaltasks(e.g.,countingbackward)tofacilitate cant,andthefollowingresultsrefertoalineareffect.
a return to baseline (Chivers et al., 2004). In the other dataset Lesbians(Kinseyscores5–6)respondednonsignificantlymore
participants watched, in random order, sexual stimuli alternating to same-sex stimuli as compared with straight women (Kinsey
withnaturescenesthatfacilitatedareturntobaseline(Riegeretal., scores 0–1), p (cid:4) .13, (cid:6) (cid:4).12 [(cid:3).04, .28] (Figure 1A) and
2015). Finally, participants completed a questionnaire with demo- significantly less to other-sex stimuli, p (cid:4) .05, (cid:6) (cid:4) (cid:3).16
graphic information, sexual orientation, and masculinity-femininity [(cid:3).32,(cid:3).01](Figure1B).Wethencalculatedadifferencescoreof
andwerepaid($50or$100,dependingonthedataset).Procedures genital response to same-sex versus other-sex stimuli. For this
tookapproximately120min. difference, straight women responded similarly to the sexes (a
Figure 1. Women’s responses to sexual stimuli. Reported sexual orientation of 152 women in relation to
genitalarousaltothesamesex(A),othersex(B),andtheirdifference(C).OntheYaxes,genitalarousalscores
reflect changes compared to baseline, z-scored within participants. On the X axes, 0 represents exclusively
straight, 3 bisexual, and 6 exclusively lesbian. Lines represent regression coefficients with 95% confidence
intervals.Dotsrepresentparticipants’averagescores.Statisticsrepresentlineareffects.
SEXUALAROUSALANDMASCULINITY-FEMININITY 271
score of 0), whereas lesbians responded stronger to the same sex remained similar in magnitude before and after including
thantheothersex,p(cid:4).01,(cid:6)(cid:4).20[.04,.35](Figure1C). masculinity-femininityasacovariate(seeFigure1andTable2).In
Hypothesis 2. We regressed reported adulthood masculinity- fact, a comparison among only those women who reported their
femininity onto women’s sexual orientation. Consistent with the masculinity-femininitysuggestedthattheeffectofsexualorienta-
hypothesis, lesbians were more masculine than feminine, com- tion on sexual arousal to the same or other sex increased after
pared with straight women, p (cid:2) .0001, (cid:6) (cid:4).42, [.25, .56]. In includingmasculinity-femininityasacovariate,p(cid:4).03,(cid:6)(cid:4).20
Figure2,the95%confidenceintervalsoftheregressioncoefficient [.02,.37],andp(cid:4).003,(cid:6)(cid:4).30[.08,.52],respectively.
show that straight women were below the midpoint of possible We then tested systematically whether masculinity-femininity
masculinity-femininityscores(ascoreof4)whereaslesbianswere mediatedtherelationofsexualorientationwithsexualarousalby
justabove. computingmediationanalysesonthebasisof1,000bootstrapped
Acrosswomenofallsexualorientationstherewasvisiblevari- samples (Preacher & Hayes, 2008). Given the distribution of its
ation in masculinity-femininity scores (see Figure 2). It was pos- confidence intervals, the indirect effect of sexual orientation on
sible,though,thatlesbiansvarymorestronglythanotherwomen sexualarousaltothesamesex(i.e.,theportionofthiseffectthat
(Lippa, 2005b). However, a Levene test for unequal variance is influenced by masculinity-femininity) differed significantly
(whichcomparesthemagnitudeofabsoluteresidualsacrosssexual fromzero,(cid:6)(cid:4)(cid:3).14[(cid:3).24,(cid:3).06].Fromthecomparisonofeffect
orientations)didnotsupportthis,p(cid:4).14,(cid:6)(cid:4)(cid:3).13[(cid:3).31,.06]. sizes in the last paragraph, this meant that controlling for
Hence,thevariationofmasculinity-femininitywassimilaracross masculinity-femininitysignificantlyincreasedthiseffectofsexual
sexualorientations. orientation. This suggests “suppression” rather than mediation;
Hypotheses 3. We hypothesized if lesbians are more male- that is, the predictive power of sexual orientation on arousal is
typicalthanstraightwomenintheirsexualarousalandnonsexual weakened in the absence of masculinity-femininity (MacKinnon,
behaviors,thentherelationshipoffemalesexualorientationwith Krull,&Lockwood,2000).
sexualarousalwouldbemediatedbynonsexualmasculinebehav- Notably,higherdegreesofadulthoodmasculinitypredictedless
iors.Weconductedmultipleregressionanalysespredictinggenital genital arousal to the same sex, regardless of sexual orientation
arousal to the same sex, other sex, and their difference score. (seeTable2).Thiswasnotanartifactduetocollinearitybecause
Independent variables were sexual orientation and self-reported of the correlation of masculinity-femininity with sexual orienta-
adulthood masculinity-femininity. If male-typical sexual arousal tion. The simple relationship of genital response to the same sex
patterns of lesbians were linked to their nonsexual masculinity, withadulthoodmasculinity-femininitywasalsonegative,p(cid:4).02,
then the inclusion of self-reported masculinity-femininity as an (cid:6)(cid:4)(cid:3).23[(cid:3).39,(cid:3).04].Wehadnospecifichypothesisaboutthis
independentvariableshoulddecreasetherelationofsexualorien- pattern,anditisunclearwhetheritismeaningful.
tationwithsexualarousalpatterns. Based on further mediation analyses with bootstrapping, the
Table 2 summarizes the results of the analyses that included effectofsexualorientationonsexualarousaltotheothersexwas
bothindependentvariables.Themaineffectsofsexualorientation not significantly mediated by masculinity-femininity, (cid:6) (cid:4) (cid:3).02
[(cid:3).08, .07]. Similar the above analyses, the difference in sexual
arousaltothesamesexandothersexwassignificantlyenhanced
inthepresenceofmasculinity-femininity,(cid:6)(cid:4)(cid:3).09[(cid:3).21,(cid:3).02].
Hypotheses 4. Alternatively to Hypothesis 3, we hypothe-
sizedthatmasculinity-femininitywouldmoderatetherelationship
of sexual orientation with sexual arousal. The most masculine-
behavinglesbianswouldbemostgenitallyarousedtothesamesex
than the other sex, in comparison with both straight women and
less masculine lesbians. We conducted three multiple regression
analysespredictinggenitalarousaltothesamesex,othersex,and
theirdifferencescore.Independentvariablesweresexualorienta-
tion,masculinity-femininity,andtheirinteraction.Ifmasculinity-
femininity differentiates the genital response patterns of lesbians
more so than it does of straight women, then this interaction
between sexual orientation and masculinity-femininity will be
significant.
Results of these analyses are summarized in Table 3. For
genitalresponsetothesamesex,othersex,ortheirdifference,
there were no significant interactions between sexual orienta-
tion and masculinity-femininity. Thus, the most masculine les-
Figure2. Women’sdegreeinmasculinityandfemininity.Reportedsex-
bians did not have stronger responses to the same sex than the
ualorientationof115womeninrelationtotheirself-reportedadulthood
other sex, as compared with other women. As for the previous
masculinity-femininity. On the Y axis, a score of 7 indicates the most
analyses, the main effects of sexual orientation remained sim-
masculinescore,themiddlelineanaveragescoreof4,andascoreof1the
ilar, if not stronger, in magnitude after including masculinity-
mostfemininescore.OntheXaxes,0representsexclusivelystraight,3
bisexual,and6exclusivelylesbian.Linesrepresentregressioncoefficients femininity as a moderator (see Figure 1 and Table 3).
with95%confidenceintervals.Dotsrepresentparticipants’averagescores. Missing data. Information of self-reported masculinity-
Statisticsrepresentalineareffect. femininity was missing for 37 women. We examined whether
272 RIEGER,SAVIN-WILLIAMS,CHIVERS,ANDBAILEY
Table2
MultipleRegressionAnalysesforSexualOrientationandSelf-ReportedAdulthoodMasculinity-
FemininityPredictingGenitalArousaltotheSameSex,OtherSex,andTheirDifferenceScore
Across115Women
Responsetosame Responseto Responsetosame
sex othersex orothersex
Variables (cid:6) (cid:6) (cid:6)
Sexualorientation(SO)1 .28[.14,.43](cid:2) (cid:3).16[(cid:3).31,.01] .30[.09,.53](cid:2)
Self-reportedadulthood
masculinity-femininity(M-F)2 (cid:3).34,[(cid:3).46,(cid:3).22](cid:2) (cid:3).03[(cid:3).14,.09] (cid:3).23[(cid:3).40,.00](cid:2)
Note. R2’s for the three models are .12, .03, and .08, respectively. Numbers in brackets represent 95%
confidenceintervalsofthestandardizedregressioncoefficient,(cid:6).
1Higherscoresindicatestrongerorientationtothesamesexandlesstotheothersex. 2Higherscoresindicate
moremasculinityandlessfemininity.
(cid:2)p(cid:2).05.
thesewomen differed from women from whom such data were Table 3), and with imputed data, p (cid:4) .003, (cid:6) (cid:4).28 [.10, .46].
available. Multiple linear regression analyses indicated no sig- Similar to the above analyses (see Table 3), with imputed data
nificant differences between these groups. For example, the masculinity-femininityactedneitherasmediator,norasmoderator
relationship of sexual orientation with genital arousal to the (.65 (cid:2) ps (cid:2) .85, (cid:3).04 (cid:2) (cid:6)s (cid:2) .04, (cid:3).20 (cid:2) CI (cid:2) .15). In total,
same sex or other sex (Figure 1C) was similar before and after analyseswithimputeddataresembledanalysesdescribedabove.
controlling for a variable that specified available or missing Covariates. Lesbians were on average older and more edu-
information on masculinity-femininity, p (cid:4) .01, (cid:6) (cid:4).20 [.04, cated than other women (see Table 1). Although sexual orienta-
.35],andp(cid:4).02,(cid:6)(cid:4).24[.05,.47],respectively.Furthermore,
tions did not significantly differ by the proportion of being Cau-
this relationship was not significantly moderated by whether
casian,itwaspossiblethatthisvariable,too,hadaneffectonthe
information on masculinity-femininity was missing, p (cid:4) .78,
dependentmeasurers.Weconductedregressionanalysessimilarto
(cid:6) (cid:4).03 [(cid:3) .18, (cid:3).24].
those described above (see Table 3), but included age, education
Because sexual orientation was a predictor of masculinity-
level (scored continuously), and ethnicity (Caucasian or other
femininity(seeFigure2),wecomputedmultipleimputations(Lit-
ethnicity)ascovariates.Themaineffectsofsexualorientationon
tle & Rubin, 2002) to estimate missing values of masculinity-
genitalarousalpatternswerecomparableinmagnitudebeforeand
femininityfromitscovariancewithsexualorientation.Acrossfive
aftercontrollingforthesevariables.Forexample,themaineffect
imputations, the pooled effect of sexual orientation with self-
ofsexualorientationonthedifferenceinarousaltothesamesex
reported masculinity-femininity was similar to the effect with
list-wiseexclusionsofmissingdata,p(cid:2).0001,(cid:6)(cid:4).39[.20,.60], andothersexremainedsimilar,p(cid:4).003,(cid:6)(cid:4).30[.08,.52](see
andp(cid:2).0001,(cid:6)(cid:4).42[.25,.56],respectively.Moreover,effects Table 3), and p (cid:4) .002, (cid:6) (cid:4).34 [.12, .58], respectively. These
on sexual arousal were comparable for analyses with list-wise main effects (or their interactions with masculinity-femininity)
excluded data and imputed data. For example, the relation of were not significantly moderated by age, ethnicity, or education
sexualorientationwithgenitalarousaltothesamesexorothersex (.11 (cid:2) ps (cid:2) .99, (cid:3).09 (cid:2) (cid:6)s (cid:2) .17, (cid:3).23 (cid:2) CI (cid:2) .31). Thus,
(with masculinity-femininity as a covariate) was small to modest assessed demographic variables had little effect on the link of
in effect with excluded data, p (cid:4) .003, (cid:6) (cid:4).30 [.08, .52] (see sexualorientationwithmasculinity-femininityandsexualarousal.
Table3
MultipleRegressionAnalysesForSexualOrientationandSelf-ReportedAdulthoodMasculinity-
FemininityPredictingGenitalArousaltotheSameSex,OtherSex,andTheirDifferenceScore
Across115Women
Responsetosame Responseto Responsetosame
sex othersex orothersex
Variables (cid:6) (cid:6) (cid:6)
Sexualorientation(SO)1 .28[.12,.43](cid:2) (cid:3).17[(cid:3).33,.02] .30[.08,.52](cid:2)
Self-reportedadulthood
masculinity-femininity(M-F)2 (cid:3).34,[(cid:3).46,(cid:3).21](cid:2) (cid:3).02[(cid:3).14,.10] (cid:3).23[(cid:3).41,.00](cid:2)
SO(cid:7)M-F (cid:3).00[(cid:3).13,.10] .04[(cid:3).08,.15] (cid:3).03[(cid:3).20,.15]
Note. R2’s for the three models are .12, .03, and .08, respectively. Numbers in brackets represent 95%
confidenceintervalsofthestandardizedregressioncoefficient,(cid:6).
1Higherscoresindicatestrongerorientationtothesamesexandlesstotheothersex. 2Higherscoresindicate
moremasculinityandlessfemininity.
(cid:2)p(cid:2).05.
SEXUALAROUSALANDMASCULINITY-FEMININITY 273
Alternative analyses. Because Hypotheses 3 and 4 were not en’s dilation patterns to sexual stimuli. These datasets differed
confirmed, we investigated whether alternative analyses would somewhatintheirmethodologies,asdescribedbelow.Differences
give more informative results. One way of analyzing data is to did not statistically affect the relationships of sexual orientation
predictwomen’ssexualorientationidentities(seeTable1)bytheir withpupildilationormeasuresofmasculinity-femininity.
genital responses to males and females, as it is the case with
discriminant analyses (rather than vice versa, as for the reported Method
regressionanalyses).Whendiscriminantanalyseswereconducted,
thesexualorientationsofthemajorityofwomen(82%)werenot Participants. Advertisementswereplacedonwebsitesinan
successfullydiscriminatedbasedontheirgenitalresponses.How- area close to a Northeast university. The 345 recruited women
ever, correct classification was significantly stronger for lesbians self-identifiedas“straight”(n(cid:4)68),“mostlystraight”(n(cid:4)63),
thanforotherwomen(seeTable4).Alogisticregressionanalysis “bisexualleaningstraight”(n(cid:4)34),“bisexual”(n(cid:4)31),“bisex-
confirmedthatcorrectclassificationsweresignificantlymorecom- ual leaning lesbian” (n (cid:4) 43), “mostly lesbian” (n (cid:4) 66), and
mon for lesbians than other women, B (cid:4) 1.63 [1.00, 2.41], p (cid:2) “lesbian”(n(cid:4)40).Theaverageage(SD)was23.03(5.47)years.
.0001,OD(cid:4)26.10[7.32,124.37].Thesefindingscomplementthe The most common ethnicity was Caucasian (65%), followed by
reported results for Hypothesis 1, with lesbians being more mixedethnicities(11%),Black(7%),andHispanic(6%).Educa-
strongly arousal to the same sex, whereas other women had less tionwascodedas1(nohighschool),2(somehighschool),3(high
distinct arousal to either sex (see Figure 1). Yet, similar to the school diploma), 4 (some college), 5 (college graduate), and 6
aforementionedanalyses,sexualorientationdifferencesincorrect (postgraduatestudentordegree).Theaverageeducation(SD)was
classifications,basedonarousal,wereneithermediatednormod- 4.57 (.84); the median was 4.00. Most common was “some col-
eratedbytheirmasculinity-femininity. lege”with57%;alloftheseparticipantswerecurrentlyincollege.
Insum,lesbiansweresomewhatmorearousedtothesamesex Table 5 shows that lesbians were significantly older than other
than the other sex, consistent with previous research (Chivers et sexualorientationgroups,p(cid:4).0002,R2(cid:4).07,hadmoreeduca-
al.,2007),andreportedmoreadulthoodmasculinitythanstraight tion,p(cid:4).01,R2(cid:4).05,andhigherproportionofbeingCaucasian,
women,alsoconsistentwithpreviouswork(Lippa,2008b).How- (cid:5)2(6) (cid:4) 20.1, p (cid:4) .003. These differences did not significantly
ever,presentdatadidnotsupportthehypothesesthatmasculinity- affectanalyses,aswereportbelow.
femininity mediated or moderated the link of female sexual ori- Measures.
entation with sexual arousal. It is possible that Study 1 did not Sexualorientation. Participantsreportedtheirsexualorienta-
employthemosteffectivemeasurestoelicitthepredictedeffects. tion identities, attractions, and fantasies in adulthood on Kinsey-
Therelationshipoffemalesexualorientationwithsexualresponse typescales(Kinseyetal.,1953).Measureswerehighlycorrelated
can be stronger if measured by pupil dilation rather than genital (ps(cid:2).0001,.82(cid:2)rs(cid:2).94,.78(cid:2)95%CI(cid:2).95),andaveraged
arousal(Rieger&Savin-Williams,2012a).Perhaps,pupildilation within participants. For this composite, a score of 0 represented
isalsothemoreprecisemeasureforassessinganyeffectsofsexual exclusivelystraightandascoreof6exclusivelylesbian.
orientationandmasculinity-femininityonfemalesexualresponse. Masculinity-femininity. Reported childhood masculinity-
Furthermore, the relationship of masculinity-femininity with sex- femininity was assessed in 186 women (Rieger et al., 2015; un-
ual orientation can be stronger if masculinity-femininity is mea- publisheddata)withtheChildhoodGenderIdentityScale(Rieger
suredwithself-reportsaboutchildhoodorwithobservationsfrom et al., 2008). Statements included “I was a masculine girl,” and
adulthoodbehaviors,ratherthanwithself-reportsaboutadulthood “As a child I preferred playing with boys rather than girls,” and
characteristics (Rieger et al., 2008; Rieger et al., 2010; Rieger & were endorsed with 7-point scales. Reported adulthood
Savin-Williams,2012b).Perhaps,too,thesemeasuresaresuperior masculinity-femininity was measured in all 345 women with the
for assessing effects of sexual orientation and masculinity- Continuous Gender Identity Scale (Rieger et al., 2008) described
femininityonsexualarousal.Study2investigatedthesepossibil- in Study 1. Cronbach’s alpha exceeded .85 for each scale. Two
itieswithasamplelargerthanforStudy1,therebyincreasingits averages were computed for each participant, one each for self-
powerfordetectingpotentialeffects. reportedchildhoodandadulthoodmasculinity-femininity.
Across two datasets (Rieger & Savin-Williams, 2012a; unpub-
lished data) adulthood masculinity-femininity of 273 participants
Study 2
wasassessedwithobserver-ratings.Eightyundergraduatestudents
Study 2 combined two datasets (Rieger et al., 2015; Rieger & (20 straight and gay men and women) viewed, in random order,
Savin-Williams, 2012a) with unpublished data to examine wom- 10-s videos of participants discussing the weather. Raters can
Table4
DistributionofCorrectlyandIncorrectlyClassifiedSexualOrientationIdentitiesBasedonWomen’sGenitalResponsestotheSame
SexandtheOtherSex
Bisexualleaning Bisexualleaning
Women(N(cid:4)152) Straight Mostlystraight straight Bisexual lesbian Mostlylesbian Lesbian
Correct 8(26) 2(05) 2(14) 1(10) 2(11) 1(05) 15(78)
Incorrect 23(74) 39(95) 12(86) 9(90) 16(89) 18(95) 4(22)
Note. Numbersinbracketsarepercentagescores.
274 RIEGER,SAVIN-WILLIAMS,CHIVERS,ANDBAILEY
Table5
DistributionofSexualOrientationIdentitiesAcrossAges,Ethnicities,andEducationinStudy2
Bisexualleaning Bisexualleaning
Women(N(cid:4)345) Straight Mostlystraight straight Bisexual lesbian Mostlylesbian Lesbian
Number 68 63 34 31 43 66 40
Averageage 21.62[20.36,22.87] 21.57[20.70,22.44] 22.97[21.75,24.19] 24.16[21.70,26.62] 21.86[21.01,22.71] 24.88[22.97,26.79] 25.08[23.46,26.69]
PercentageCaucasian 59[47,70] 68[56,78] 47[31,63] 68[50,81] 63[48,76] 82[71,89] 83[68,91]
Averageeducation1 4.13[3.94,4.32] 4.46[4.26,4.66] 4.53[4.25,4.80] 4.61[4.20,5.02] 4.37[4.12,4.62] 4.65[4.46,4.84] 4.60[4.33,4.87]
Note. Numbersinbracketsrepresent95%confidenceintervals.
1Educationwascodedas1(nohighschool),2(somehighschool),3(highschooldiploma),4(somecollege),5(collegegraduate),6(postgraduatestudent
ordegree).
reliablyassessmasculineandfemininebehaviorsandappearance Pupil data. Pupil dilation to sexual stimuli reflects sex and
fromsuchbriefvideos(Riegeretal.,2010).Inonedataset(Rieger sexualorientationdifferencesingenitalarousal,suggestingitisa
& Savin-Williams, 2012a), raters used a 7-point scale for their validindicatorofsexualresponse(Hess,Seltzer,&Shlien,1965;
assessments, ranging from 1 (very feminine) to 4 (average) to 7 Rieger et al., 2015; Rieger & Savin-Williams, 2012a). A SR
(verymasculine).Intheunpublisheddata,ratersusedthreescales ResearchRemoteinfraredgazetrackerrecordedpupildataevery2
askinghowmasculine,feminine,andhowmasculineversusfem- mswitha16mmlens(Rieger&Savin-Williams,2012a)orevery
inineparticipantswere.Thesescalesrangedfrom1(notatall)to millisecond with a 35 mm lens (Rieger et al., 2015; unpublished
4(average)to7(verymuch). data). Lenses were focused on participants’ preferred eye. The
In each dataset and for each rater group, ratings on the same program EyeLink computed pupil area as the number of the
scalewerereliable(allCronbach’s(cid:8)s(cid:2).94).Ratingsofthefour tracker’scamerapixelsoccludedbytheinfraredlightreflectedby
groupswerehighlycorrelated(ps(cid:2).0001,.80(cid:2)rs(cid:2).97,.71(cid:2) thepupil.Ifpupilsdilatedwhileviewingstimuli,morepixelswere
95%CI(cid:2).98);therefore,theirratingswerecombinedforfurther occluded.Becauserawpupilareadataincluded“0’s”formissing
analyses.Inonedataset(Rieger&Savin-Williams,2012a),ratings values,forexamplefromblinksorheadmovements,thesevalues
were averaged within participant to obtain an overall score with wereremovedpriortoanalyses.
higher numbers indicating more observer-rated masculinity than Procedure. Participants provided written informed consent
femininity. In the unpublished data, the reverse of participants’ andwereseatedinadimlylitroomfacingascreenwithresolution
average femininity scores correlated strongly with their average of1,024by768pixels.Participants’headsrestedonamount500
masculinity scores and their masculinity-femininity scores, ps (cid:2) mm from the eye tracker’s lens. For calibration of pupil data,
.0001, .97 (cid:2) rs (cid:2) .99, .95 (cid:2) 95% CI (cid:2) .99. Thus, these three participants fixated and refixated their gaze on nine points that
scores were further averaged to compute an overall score with defined the outline of the screen. Participants were instructed to
higher numbers indicating more observer-rated masculinity than watch all videos carefully, regardless of whether they liked the
femininity. content. First, participants watched one of the neutral stimuli
Acrossavailabledata,thethreemeasuresofmasculinity-femininity (landscapes or cloud animations), followed, in random order, by
(self-reportfromchildhoodandadulthood,andobserver-ratingsfrom presentations of sexual stimuli, alternating with questions about
adulthood) were modestly correlated (ps (cid:2) .0001, .44 (cid:2) rs (cid:2) .54, thevideos(Rieger&Savin-Williams,2012a)oracombinationof
.28(cid:2)CI(cid:2).64).However,onedatasetdidnotincludeinformationon questionsandnaturescenes(Riegeretal.,2015;unpublisheddata).
self-reported childhood masculinity-femininity (Rieger & Savin- Thelastvideowasthefinalneutralstimulus.
Williams,2012a),whereasoneotherdidnothavedataonobserver- After watching stimuli privately, 273 participants were vid-
ratedadulthoodmasculinity-femininity(Riegeretal.,2015).Thiswas eotaped, seated in a chair and discussing winters in the North-
the case because these projects have been independently designed eastU.S.(Rieger&Savin-Williams,2012a;unpublisheddata).
withdifferentemphasizesgiventothesemeasures.Belowwereport The first full sentence given as an answer within the first 10 s
analysesthatinvestigatedwhethermissingdataaffectedtherelation- wasusedforsubsequentobserver-ratingsofparticipant’sadult-
shipsofsexualorientation,masculinity-femininity,andpupildilation. hood masculinity-femininity. Finally, participants completed
Stimuli. Inonedataset(Rieger&Savin-Williams,2012a)and questionnaires about demographics, sexual orientation, and
theunpublisheddata,30-svideosof12malestimuliand12female masculinity-femininity and received payment ($30 or $100,
stimulishowedeitheranakedmaleorfemalemodelmasturbating. depending on the dataset). The procedure took 45 to 120 min,
Thirty-secondvideosofnaturescenes(landscapesoranimationsof depending on the dataset.
clouds)wereusedasneutralstimuli.Intheotherdataset(Riegeret There is no consensus as to the most appropriate technique of
al., 2015), 3-min videos of three male stimuli and three female analyzing pupil size data (Otero, Weekes, & Hutton, 2011). We
stimuli were used as sexual stimuli and 2-min videos of cloud decidedonproceduresidenticaltothoserecommendedforanalyz-
animations for neutral stimuli. This difference in stimulus length inggenitalarousalresponses(Harrisetal.,1992).Pupildatawere
(30svs.severalminutes)didnotaffectresults.Allstimuliwereof averaged within stimuli and for each participant. Because pupils
similarluminance;furthermore,luminancewassettoequalupper varyinsizeanddegreeofdilation,theseaverageswere,equivalent
andlowerthresholdsacrossstimulibyusingtheprogramsMPEG to the genial arousal data, z-scored within participants. Average
Streamclip and Final Cut Pro. Videos had a resolution of 768 by dilationtoneutralstimuliwassubtractedfromtheaveragedilation
536pixelsandwerepresentedfullscreen. to each sexual stimulus. We then computed, for each participant,