<<

Arch Sex Behav (2012) 41:467–476 DOI 10.1007/s10508-011-9785-6

ORIGINAL PAPER

The Partner-Specific Sexual Liking and Sexual Wanting Scale: Psychometric Properties

Tamar Krishnamurti • George Loewenstein

Received: 9 March 2010 / Revised: 6 December 2010 / Accepted: 21 March 2011 / Published online: 1 July 2011 Ó Springer Science+Business Media, LLC 2011

Abstract Inspired by research showing that wanting (one’s Keywords Sexual liking Á Sexual wanting Á motivation to engage in an activity) often diverges from liking Intimate relationships Á Scale development (one’s enjoyment of the activity), this article details the devel- opment and validation of a new measure to examine the dis- tinction between sexual wanting and liking within a relation- Introduction ship: the partner-specific sexual liking and wanting (PSSLW) scale. In Study 1, participants (N = 1145; 63% female) com- Logically, one might think that wanting to engage in an activity pleted items intended to measure PSSLW. Factor analysis sup- should be virtually indistinguishable from the liking of that activ- ported a 15-item two-factor solution that explained 64.7% of ity, i.e., how much we anticipate enjoying it. Yet, according to a the total variance. The partner-specific sexual liking (PSSL) provocative hypothesis advanced by Berridge (see Smith & Ber- subscale (Cronbach’s a = .93) and the partner-specific sexual ridge, 2007; Treit & Berridge, 1990;Wyvell&Berridge,2000), wanting (PSSW) subscale (Cronbach’s a = .87) showed good wanting and liking do not necessarily coincide. Indeed, a number internal validity. Test–retest reliability on a subsample (n = 30) of studies suggest that the experiences of liking and wanting are was high (Pearson’s r = .75). In Study 2, participants (N = 67; processed in overlapping, but somewhat distinct regions of the 71.6% female) completed the PSSLW scale and additional brain, strengthening the claim that, at least in some contexts, lik- measures of satisfaction and desire. Both scales displayed sat- ing and wanting should be represented as unique constructs (Pec- isfactory discriminant and convergent validity. In Study 3, parti- in˜a, Berridge, & Parker, 1997). cipants (N = 2589; 45.3% female) completed the PSSLW scale Deviations between wanting (the motivation to engage in an and answered questions about sexual behavior within their rela- activity) and liking (the enjoyment of the activity) have been tionships. The two subscales were distinctly correlated with observed in domains as diverse as alcohol consumption (Hobbs, measures of self-reported behavior. Moreover, confirmatory fac- Remington, & Glautier, 2005), smoking (Brauer, Cramblett, tor analysis (CFA) yielded a good-fit two-factor model, where Paxton,& Rose, 2001),druguse(Brauer& de Wit,1997; Breiter the Comparative Fit Index (CFI) = .97, Tucker Lewis Index et al., 1997), and food consumption (Finlayson, King, & Blun- (TLI) = .96, and root mean square error of approximation dell, 2007). For example, Hobbs et al. (2005) administered var- (RMSEA) = .06. Data from these three studies suggested that ious alcoholic andnon-alcoholic beverages to heavyversus light PSSLW were distinct, measurable, and valid constructs that drinkers. Even though heavy drinkers displayed higher levels of have thepotentialtoenrich future studies of sexualexperience wanting alcohol, as manifested by higher levels of consumption, and behavior within sexual partnerships. there were no significant differences in ratings of enjoyment of the drinks. In research examining the distinction between liking and wanting food, no significant difference was found in the pleasantness rating of foods between obese and lean subjects, although obese subjects consumed significantly more calories & T. Krishnamurti ( ) Á G. Loewenstein (Cox, Perry, Moore, Vallis, & Mela, 1999). Other studies have Department of Social and Decision Sciences, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, USA shown that palatability of food is not consistently correlated e-mail: [email protected] with intake (Lucas & Bellisle, 1987). While the domains in 123 468 Arch Sex Behav (2012) 41:467–476 which liking and wanting have been shown to differ are pri- Previous work on sexual quality has examined general marily areas of hedonic consumption, this distinction has not yet sexual satisfaction, which often encompasses (and confounds) been quantifiedin the highlyhedonic domainof sexual behavior. sexual frequency, sexual liking, and sexual wanting. Moreover, There are a variety of interesting patterns with respect to to the extent that prior scales have examined sexual wanting, sexual behavior that might be better understood by making a they tend to address sexual desire as an innate force that is not distinction between liking and wanting for sex. Specifically, specific to individual partners, but rather that motivates sexual gender differences in sexual initiation and reported enjoyment behavior more generally (Krafft-Ebing, 1886/1965). Here, we of sexual activity may be related to differences in levels of draw a distinction between, and seek to independently measure, sexual liking and sexual wanting between men and women. In partner-specific sexual liking (PSSL), which we define as the one study, sex was rated as the single activity, which produced hedonic experience specific to sexual interaction with a specific the greatest amount of happiness in women’s lives (Kahneman, partner, and partner-specific sexual liking (PSSW), which we Krueger, Schkade, Schwarz, & Stone,2004). Moreover, women define as the motivation to engage in sexual activity with that have a greater physiological capacity to enjoy sex than men do partner. (Baumeister, Catanese, & Vohs, 2001). Yet, women act as the Creating a scale that can measure sexual wanting for both sexual ‘‘gate keepers’’ within a relationship, determining when sexes is challenging because, as argued by Toates (2009), sexual and whether sexual contact will occur (Cohen & Shotland, wanting may operate somewhat differently in men and woman. 1996; Meston, Trapnell, & Gorzalka, 1998). This pattern of Toates found that women’s wanting may be more closely women limiting potentially highly enjoyable sexual contact, associated with deliberate and controlled processes. Men’s even after the eve of the feminist movement and more freely wanting, on the other hand, may be more closely associated with available contraceptives, could be explained if women ‘like’ sex automatic processes and thus less susceptible to factors like the more than men do, but ‘want’ it less. social context of a sexual interaction. Therefore, it was impor- In addition to gender differences, sexual frequency has been tant to ensure that our scale items did not have strong social or shown to be susceptible to influences of time, with patterns of cultural factors embedded in the text, which may influence decline in sexual frequency occurring with both age and rela- women’s positive and negative cognitions of those items. For tionship duration (Call, Sprecher, & Schwartz, 1995;Rao& example, both men and women have been shown to experience DeMaris, 1995). While some of this decrease may be due to comparable levels of in response to pornographic changes in health status or background variables (such as work material (Fisher & Byrne, 1978), yet social norms may make schedule), these variables cannot fully account for this decline women feel less comfortable actively seeking out pornographic (Greenblat, 1983). Whether the remaining diminishment in stimuli. As a result, items such as frequency of use sexual activity over time together is due to a lack of enjoyment with one’s partner would serve as a poor measure for sexual or a lack of desire or both is not well understood. wanting and, consequently, we avoided including items in our Lastly, sexual frequency has been closely tied to happiness scale that could have a socially or culturally normative inter- across genders, with one study showing that increasing a per- pretation. son’s sexual activity from once a month to once a week would be In addition, while men and women can both recognize key approximately equivalent in terms of happiness to receiving a physiological indicators of their desire, such as genital arousal, pay increase of $50,000 (Blanchflower & Oswald, 2004). some researchers (e.g., Chivers, Rieger, Latty, & Bailey, 2004; However, studies that examine patterns of sexual frequency do Chivers, Seto, Lalumie`re,Laan,&Grimbos,2010;Laan& not generally distinguish between sexual liking and sexual Everaerd, 1995) have inferred from the lower correlation wanting. If issohighlyvaluedasasourceof between self-report and the onset of these physiological indi- happiness, it is crucial to determine what aspects of intercourse cators amongst females, that women show a lack of awareness are actually impacting individuals. It may be the general of physical arousal relative to men. To account for this possi- wanting of sex and the carrying out of those desires that is bility, we included items in our wanting scale that were intended beneficial. Alternately, it may be the more subtle distinction of to capture both automatic physiological arousal responses, as liking the sex that we have, which has a positive impact on our well as more subjective assessments of desire. well-being. A similar challenge exists with respect to measuring sexual To understand changes in sexual initiation and frequency liking due to the common observation that men and women tend with age, relationship duration, and gender (barring changes in to derive maximal enjoyment from different types of activities physical ability to engage in sex over time), it is critical to (Baumeister, 2000; Chivers & Bailey, 2004). To avoid designing distinguish between the desire for and the anticipated enjoyment a measure that yielded spurious differences based on the match of sex. Yet, until now, the psychological distinction between between items included in the scale and the different specific liking and wanting for sex within a relationship has not been predilections of the two genders, our measure of liking did not explicitly measured with a single scale. focus on the pleasure gained from specific sexual activities.

123 Arch Sex Behav (2012) 41:467–476 469

In what follows, we present findings from a series of studies Carey, & Steinberg, 1996) or generated by the authors to mea- that were conducted to develop and validate a scale comprised sure dimensions of PSSL and PSSW that they perceived to be of two subscales, which assess sexual liking and wanting for a missing from existing scales (see Tables 1, 2). Specifically, sexual partner. In Study 1, factor analysis was conducted on a Items 1, 2, 5, and 6 were drawn directly from the instruments series of items in order to identify two factors, which measure mentioned above. Items 3, 4, 7, and 11 were modified from the sexual liking and wanting for a partner. Test–retest reliability of instruments mentioned above to capture the distinction between these two factors was measured on a subsample of participants. sexual liking and sexual wanting within a specific relationship. In Study 2, discriminant and convergent validity were examined The remaining scale items were created to capture dimensions by administering the two sexual liking and wanting subscales, as of liking and wanting not incorporated in existing scales. well as established measures of satisfaction and desire. To Participants rated the frequency with which they experienced examine the relationship between the subscales and real behav- the thought, feeling or behavior captured by each item. Partic- ior, participants in Study 3 were asked to complete the sexual ipantsrespondedtoItems1–10andItems16–22ona5-point liking and wanting subscales, and answered questions about scale with anchors ranging from 1 = Rarely or never to 5 = sexual behavior within their relationships. A confirmatory fac- Always. Item 11 was measured on an 8-point scale and Item 12 tor analysis (CFA) was also conducted. was measured on a 9-point scale. Item 11 measured frequency of sexual thoughts, broken down into eight ordered frequency categories, ranging from ‘‘1 = notatall’’to‘‘8= many times a Study 1: Item Selection day.’’Item 12 measured degree of intensity of those thoughts and had an obvious midpoint. The scale ranged from‘‘1 = not at Method all strong’’to‘‘9 = extremely strong.’’

Participants Analysis

Participants were 1,145 volunteers recruited from advertise- Principal components analysis (PCA) was conducted to assess ments posted on the electronic bulletin board www.craigs the underlying factor structure of the administered items. PCA list.org in major cities across the United States and through a link was chosen to identify a unique factor solution as it avoids the placed on a New York Times recruitment website. Participants factor indeterminacy issues of alternate factor methods, such as were eligible to participate if they were 18 years or older and exploratory factor analysis (Elffers, Bethlehem, & Gill, 1978). identified themselves as being in a sexually active relationship. It was anticipated that factors would be correlated with one Upon completing the questionnaire, participants were given the another; therefore, an oblique rotation procedure (direct obli- opportunity to enter a lottery with a 1/50 chance of winning a min) was used. Items with primary factor loadings C.40 and $50 gift certificate (doing so required providing an e-mail secondary factor loadings B.30 were retained. Items that did not address, which not all chose to do). All participants completed load on more than one factor were also retained. Items that did measures online. The questionnaire received University Institu- not meet these criteria were removed until all items met reten- tional Review Board approval and, due to its anonymous nature, tion criteria. did not require informed consent. Participants were excluded from analyses if they were not in a heterosexual relationship Results (n = 18). Women comprised 63% of the sample. Participants ranged The majority of the items (18 out of 22) had a high correlation (.3 inagefrom18to98(Myears = 31.91, SD = 11.20). Length of or greater) with at least one other item. The Kaiser–Meyer–Olkin participants’ primary romantic relationship ranged from less than measure of sampling adequacy of these items was .94, much a month to 52 years (Mmonths = 83.51, SD =96.34). The racial above the recommended value of .6, and Bartlett’s test of sphe- distribution of the sample was 81% Caucasian, 2.5% African- ricity was significant, v2(231) = 13068.23, p\.001. The results American, 8.9% Asian, and 3.2% Hispanic. The breakdown of of these tests suggested that the variables were likely to factor the sample on highest level of education completed was: less than well, based on their correlations and partial correlations. Com- college 9.9%, some college or a college degree 66.9%, and post- munalities were all above .3, confirming the common variance of graduate degree 23.1%. each item. After several iterations, in which four items were eliminated because they had low inter-item correlation and three Measures items were removed because they did not contribute to a simple factor structure, a total of 15 items were retained. Participants completed 22 items either drawn and modified The best-fit solution revealed two components with eigen- from existing measures of general sexual satisfaction (Hudson, values[3.0. These two components explained 64.7% of the var- Harrison, & Crosscup, 1981) and general sexual (Spector, iance. The first component, which we labeled‘‘partner-specific 123 470 Arch Sex Behav (2012) 41:467–476

Table 1 Items and factor loadings for the PSSL and PSSW scales Item Component 1 Component 2 Communalities

Partner-specific sexual liking Item 1: My partner is sexually very exciting .74 .17 .72 Item 2: Sex is fun for my partner and me .78 .11 .71 Item 3: Our sexual relationship lacks qualitya .82 -.18 .54 Item 4: Sex with my partner is wonderful .80 .14 .77 Item 5: My partner is very sensitive to my sexual needs and desires .84 -.07 .65 Item 6: Our is very exciting .82 .09 .76 Item 7: I feel that our sex life is boringa .83 -.12 .60 Item 8: I enjoy the techniques my partner likes or use .79 .04 .65 Item 9: I lose track of time when I have sex with my partner .55 .20 .46 Item 10: My sexual fantasies feature my partner .63 .17 .54 Partner-specific sexual wanting Item 11: Thinking about the last month, how often have you had sexual thoughts .21 .58 .51 about your primary sexual partner when you were not engaging in sexual activity?a Item 12: When you have sexual thoughts about your primary sexual partner, .23 .65 .63 how would you rate the intensity of those feelings? Item 13: When you look at your primary sexual partner, how often does this .00 .80 .78 result in physical sexual arousal (e.g., an , increased heart rate, lubrication, etc.)? Item 14: When you think about your primary sexual partner, how often does this -.06 .91 .65 result in physical sexual arousal? Item 15: When you have physical contact with your primary sexual partner (e.g., hugging, -.04 .88 .74 holding hands or touching), how often does this result in physical sexual arousal? Eigenvalue 8.02 1.69 Percentage of explained variance 53.46 11.24 a Indicates a reverse-coded item

Table 2 Items and factor loadings for those items that were not retained in the PSSL and PSSW scales Item Component 1 Component 2 Communalities

Item 16: It is easy for me to get sexually excited by my partner .41 .47 .67 Item 17: I get very turned on before sex with my partner .47 .48 .64 Item 18: Our sex life is monotonousa .08 -.01 .43 Item 19: When I’m having sex with my partner, I imagine I’m having sex with someone elsea .00 .06 .38 Item 20: When we have sex, it is too hurriedly completeda .22 -.20 .43 Item 21: Sex is something that has to be endured in our relationshipa -.22 .08 .43 Item 22: During sex, I get distracted by other thoughtsa -.04 .20 .51 a Indicates a reverse-coded item sexual liking (PSSL),’’ consisted of 10 items hypothesized to the PSSW subscale, before summing the items, we calculated a measure sexual liking. The second component, ‘‘partner-spe- composite score by reweighting the individual items so that all cific sexual wanting (PSSW),’’comprised 5 items hypothesized were on a 9-point scale (e.g., so that each item was normalized to measure sexual wanting. The factor-loading matrix for this with a weight of one). The PSSL subscale items were all on a final solution is shown in Table 1. The loadings of items that 5-point Likert-type scale, so the composite score was a simple werenotretainedareshowninTable2. The inter-item corre- summation of the items. Higher scores indicated greater levels lations for retained items are shown in Table 3. of partner-specific sexual wanting and liking.1

Internal Validity and Test–Retest Reliability 1 We also created partner-specific sexual wanting and sexual liking scores in which items were weighted by factor loading. Both methods of scoring were highly correlated (r = .95 and r = .92, for wanting and liking, To create subscales, we summed the items loading on each respectively). We also determined the correlation between the mean factor factor. Due to the non-uniform response scales of each item in scores for each factor and the simple summed PSSW and PSSL scores. For 123 Arch Sex Behav (2012) 41:467–476 471

Table 3 Inter-item correlation of the 15 items of the PSSL and PSSW scales Item 123456789101112131415

1 2 .69 3 .50 .56 4 .73 .77 .54 5 .63 .62 .48 .67 6 .75 .72 .54 .75 .67 7 .62 .55 .70 .58 .54 .62 8 .65 .63 .46 .70 .67 .65 .50 9 .49 .54 .37 .57 .47 .54 .39 .60 10 .59 .56 .40 .62 .55 .63 .43 .55 .47 11 .49 .47 .33 .49 .32 .44 .36 .41 .39 .47 12 .55 .53 .31 .55 .40 .54 .37 .46 .42 .51 .59 13 .46 .42 .27 .46 .33 .42 .31 .37 .38 .36 .45 .55 14 .45 .40 .22 .42 .33 .43 .26 .37 .40 .42 .52 .58 .60 15 .46 .43 .27 .46 .33 .42 .33 .36 .38 .34 .47 .54 .86 .54

Both subscales showed high internal consistency, with Table 4 Descriptive statistics for the PSSL and PSSW scales by gender Cronbach’s alphas of .87 for PSSW and .93 for PSSL. No PSSL and PSSW scales No. of Men Women increases in alpha for either scale were achievable by eliminating items M (SD) M (SD) more items. Due to concerns about differences in response by gender, data were additionally analyzed separately for men and Partner-specific sexual wantinga 5 28.62 (7.53) 27.20 (8.23) women. Both subscales of the solutionshowedhighinternal Partner-specific sexual likingb 10 35.91 (8.93) 36.53 (9.46) consistency when analyzed by gender, with a Cronbach’s alpha a PSSL scale possible scores range from 10 to 50 of .88 and .85, for PSSW in women and men, respectively, and a b PSSW scale possible scores range from 7.5 to 45 Cronbach’s alpha of .94 and .93, for PSSL in women and men, respectively. As expected, PSSL and PSSW were distinct but Study 2: Convergent and Discriminant Validity highly correlated, r = .62, with the intercorrelation between the subscales lower than their respective reliability coefficients. On The two components of the final scale of partner-specific sexual average, as predicted, men reported significantly higher levels of liking and wanting (PSSLW) showed high internal reliability, PSSW than women, t(994) = 2.69, p\.01; however, there were and the items in the PSSL and PSSW subscales showed no or no significant differences in PSSL between men and women. low cross-factor loading, indicating that the two subscales Descriptive statistics are shown in Table 4. measured distinct constructs. To test for both convergent and To assess test–retest reliability, the scale was re-administered discriminantvalidity(Campbell & Fiske,1959), weconducted a to a subsample of 30 participants 7 days after its first adminis- second study in which we administered to participants the two tration. Summed scores from the first and second administra- subscales, as well as four established related measures of sexual tions of the liking subscale correlated at r = .75. Summed scores desire, sexual satisfaction, and relationship satisfaction. from the first and second administrations of the wanting sub- scale correlated at r = .70. This degree of correlation suggested that the PSSL and PSSW scales each capture a trait that was Method relatively stable. Participants

Footnote 1 continued A second sample of 67 participants was recruited to examine the mean factor score for the liking items and the PSSL summed score, the r = .93. For the mean factor score for the wanting items and the construct validity. These participants were recruited using an PSSW summed score, the r = .94. Therefore, it is feasible to calculate online survey platform hosted by Amazon.com. Demographics partner-specific sexual liking and partner-specific sexual wanting scores were similar to the first sample, with 71.6% of the sample female in more than one way. However, for simplicity and consistency of cal- andanagerangefrom18to61(Myears = 32.91, SD = 12.14). culation by future researchers, we chose to use the simple summed scores, with each item normalized with a weight of one, for all reported Length of participants’ primary romantic relationship ranged analyses. from 2.20 months to 33.76 years (Mmonths = 87.76, SD = 123 472 Arch Sex Behav (2012) 41:467–476

105.15). The racial distribution of the sample was 76.1% Cau- both men and women, although male PSSL and male PSSW casian, 3.0% African-American, 11.9% Asian, and 7.5% His- were only weakly positively correlated with the SSI and female panic. The highest level of education obtained was: less than PSSW was only weakly positively correlated with the RAS. As college 17.9%, some college or a college degree 65.7%, and a predicted, for women, the correlations between PSSL and the post-graduate degree 16.4%. RAS and PSSL and the SSI were significantly stronger than the correlation between PSSW and the RAS, t(47) = 4.07, p\.001, and marginally stronger than the correlation between PSSW and Measures the SSI, t(36) = 1.25, p = .10. For men, however, the correla- tions between PSSL and the RAS and PSSL and the SSI were The Relationship Assessment Scale (Hendrick, 1988). The Rela- not significantly stronger than the correlation between PSSW tionship Assessment Scale (RAS) is an established measure of and the RAS or PSSW and the SSI. Conversely, the correlation generic relationship satisfaction. It consists of seven items that between PSSW and the HISD—both of which measure sexual are rated on a 7-point Likert type scale. We predicted that both desire—wasstronger than thecorrelationbetween PSSL and the wanting and liking would be moderately correlated with the RAS HISD for both men and women, although not significantly so. because sexual desire and enjoyment are both integral parts of general relationship satisfaction. However, we hypothesized that Discussion liking, which measures a form of satisfaction (sexual), would be more strongly correlated than wanting with the RAS. The PSSL and PSSW subscales were related to established The Sexual Satisfaction Inventory (Whitley & Paulson, measures of sexual desire andsatisfaction ina pattern that makes 1975). The Sexual Satisfaction Inventory (SSI) consists of 32 sense in terms of the specific content of the different scales. items and was originally designed to assess sexual satisfaction PSSW was more strongly correlated with a measure of general in married professional women. However, it has subsequently sexual desire and PSSL was more strongly correlated with mea- become an established measure of overall sexual satisfaction. sures of sexual and relationship satisfaction. We did find gender We administered this measure to ensure that our partner-specific differences in the strength of the correlations between PSSL and liking measure was adequately measuring sexual liking in satisfaction. Specifically, the correlation between PSSL and the women, as well as men. Moreover, the questions in the SSI are SSI was weaker for men than for women. The SSI measure partner-focused, making it a better validation of our measure breaks sexual satisfaction down into component sexual activi- than more general measures of sexual satisfaction. ties, such as‘‘kissing your partner’’and‘‘oral-genital stimulation Hurlbert Index of Sexual Desire (Apt & Hurlbert, 1992). This of your partner by you.’’One reason for the lack of correlation scale is a 25-item measure of general sexual desire. Scores range between sexual liking and theSSIforthemeninoursample from 0 (hyposexual desire) to 100 (hypersexual desire). This could be that male sexual satisfaction is more strongly related to scale has been shown to have predictive validity and has been satiation than enjoyment of individual activities and, as a result, used to measure sexual desire in a number of populations. Several men’s sexual satisfaction may be more strongly linked to other of the questions in the Hurlbert Index of Sexual Desire (HISD)are elements of sexual interaction, such as sexual frequency. One also partner-focused, making it a better validation of our measure reason we administered the SSI was to ensure that our partner- than more general measures of sexual drive. We hypothesized specific liking measure was correlated with sexual liking spe- that wanting, which measures partner-focused drive, would be cifically for women. The strength of the correlation between more strongly correlated with the HISD than would sexual liking. PSSL and the SSI for women suggests that this is the case. Moreover, we found that that the PSSL and PSSW subscales Results overall displayed satisfactory discriminant and convergent validity. As a final psychometric test of predictive validity, we AsshowninTable5, both the PSSW and PSSL subscales were examined the relationship between our two measures and self- positively correlated with the RAS, the SSI, and the HISD for reported sexual behavior in a third study. In Study 3, participants

Table 5 Convergent validity with three measures by gender Measures Partner-specific sexual liking Partner-specific sexual wanting Men Women Men Women

Sexual Satisfaction Inventory (SSI) .24 .61** .20 .44** Relationship Assessment Scale (RAS) .74** .66** .57* .24 Hurlbert Index of Sexual Desire (HISD) .56** .52** .64** .66** * p\.05; ** p\.01

123 Arch Sex Behav (2012) 41:467–476 473 were administered the PSSLW scale and were also asked about partner. We expected that PSSL scores would more strongly sexual initiation and enjoyment of sexual activity with their predict frequency of than PSSW scores. primary sexual partner. Many women achieve orgasm from manual stimulation, but not from intercourse alone and both men and women regularly engage in self-stimulation. As a more refined behavioral mea- Study 3: Cross Validation and Predictive Validity sure of sexual liking, we compared sexual enjoyment resulting from orgasm with one’s partner during intercourse to orgasm Method from self-stimulation. We asked participants both if they experi- enced an orgasm when having intercourse with their partner Participants (93.4% of men in our sample and 85.6% of women) and if they engagedinmasturbationtoorgasm (72.9% of men inour sample Participants for the third study consisted of a randomly selected and 55.1% of women). As a proxy for sexual liking, we com- nationally representative sample of 2,589 participants collected pared the subjective satisfaction from orgasm with partner to the through a survey research company (Survey Sampling Inter- subjective satisfaction of orgasm with self (both measured on national). Women comprised 45.3% of the sample. Participants scale from 0‘‘not at all satisfying’’to 8‘‘extremely satisfying’’) ranged in age from 18 to 88 years (Myears = 47.88, SD = 17.29). by subtracting the quality of orgasm from the Length of participants’ primary romantic relationship ranged quality oforgasm with partner. We expected PSSL scores would from less than a month to 67 years (Mmonths = 232.77, SD = be stronger than PSSW scores at predicting greater satisfaction 197.17). The racial distribution of the sample was 76.9% Cau- from partner-orgasm than masturbation-induced orgasm. casian, 14.7% African-American, 1.9% Asian, and 3.0% His- panic. The highest level of education obtained by the sample Results was as follows: less than college 27.3%, some college or a college degree 63.1%, and a post-graduate degree 9.7%. Confirmatory Factor Analysis

Cross Validation Analysis The two-factor, 15-item solution from the PCA was cross-vali- dated with this sample. Items 1–10 loaded strongly on the part- A CFA was conducted to test the validity of the two-factor ner-specific sexual liking factor and items 11–15 loaded strongly structural model derivedfrom the PCA.Todetermine the degree on the partner-specific sexual wanting factor. Yet, the initial of fit between the specified model and the sample, the following model was a poor fit. After several iterations based on modi- indiceswereused: theTuckerLewisIndex(TLI), withacutoffof fication indices, a path of covariance was then added between [.90 for an acceptable fit and[.95 for an excellent fit (Tucker & error terms for the following items: 1 and 3, 2 and 10, 3 and 10, 3 Lewis, 1973), the Comparative Fit Index (CFI), with a cutoff and 7, 7 and 10, 8 and 9, 13 and 14, 14 and 15, and 13 and 15. The of[.90 for an acceptable fit and[.95 for an excellent fit (Ben- final model shows a good fit to our data with CFI = .97, tler, 1990), and a root mean square error of approximation TLI = .96, and RMSEA = .06. The final model is shown in (RMSEA), with a cutoff of B.08 for an acceptable fit and B.05 Fig. 1. for an excellent fit (Browne & Cudeck, 1993). Modification indices made by AMOS informed path additions to the model. Predictive Validity

Predictive Validity Measures We expected that sexual wanting would be strongly associated with sexual initiation. To test this hypothesis, we regressed To illustrate the practical distinction between the partner-spe- sexual initiation on PSSL and PSSW as well as several demo- cific wanting and partner-specific liking subscales, we asked graphic variables (age, relationship duration, and gender). participants two questions about the sexual pursuit of their We also expected that sexual liking would be strongly partner, which we predicted would be affected by their levels of associated with both more frequent orgasm and a more satis- wanting. Participants were asked, ‘‘Who initiates sex?’’ There fying orgasm with one’s partner than from self-stimulation were 5 response options ranging from ‘‘I always do’’to ‘‘They (controlling for participants’ ability to achieve orgasm). To test (my partner) always do.’’ this hypothesis, we regressed the difference score of orgasm Both Hunt (1974) and Laumann, Gagnon, Michael, and satisfaction from partner and orgasm satisfaction from mastur- Michaels (1994) suggest that frequency of orgasm may be the bation on PSSL and PSSW as well as several demographic most legitimate real-world measure of sexual satisfaction, par- variables (age, relationship duration, and gender). ticularly for women. We asked participants how often they The results are shown in Table 6. The standardized regres- experienced an orgasm when having intercourse with their sion weights indicated that both PSSW and PSSL scores 123 474 Arch Sex Behav (2012) 41:467–476

Fig. 1 Final confirmatory factor analysis (CFA) model of the PSSLW scale (N = 2589)

Table 6 Regression of behavioral measures of sexual wanting and sexual liking on demographics and the PSSL and PSSW subscales Variable Who initiates sex? Orgasm frequency Difference in satisfaction between orgasm with partner and orgasm from masturbation BSEBbBSE B b BSEBb

Age .00 .00 .00 .00 .00 -.01 .00 .01 .04 Gendera -.89 .04 -.43*** .65 .06 .24*** .55 .14 .14*** Relationship duration (in years) -.01 .00 -.10** .01 .00 .07* .01 .01 .10* PSSL .02 .00 .17*** .05 .00 .35*** .07 .01 .31*** PSSW -.03 .00 -.22*** .03 .00 .17*** .01 .01 .03 F 154.24, p\.001 29.33, p\.001 R2 .25 .12 Note: Standardized coefficients are shown * p\.05; ** p\.01; *** p\.001 a For gender, male was coded as‘‘1’’and female as‘‘0’’unless specified otherwise individually predicted sexual initiation. Interestingly, they direction of the association between the initiation of sexual predicted this desire in divergent directions. Higher levels of contact with liking and wanting. PSSL were associated with higher levels of perceived partner The standardized regression weights of the orgasm fre- initiation of sexual contact and, conversely, lower levels of self- quency regression indicated that, as predicted, PSSL was more initiation. This finding suggests that partner initiation could strongly associated with frequency of orgasm than was sexual potentially play a role in altering ones’ level of sexual liking for wanting although PSSW was also a significant predictor. one’s partner. As predicted, higher levels of PSSW were asso- Additionally, the standardized regression weights of the orgasm ciated with higher levels of perceived self-initiation of sexual satisfaction regression indicated that, as predicted, PSSL was contact. Moreover, PSSW was the slightly stronger predictor of strongly associated with a more satisfying orgasm from partner the two. More research is needed to determine the causal than from self (as denoted by a positive difference score of

123 Arch Sex Behav (2012) 41:467–476 475 orgasm satisfaction from masturbation subtracted from orgasm which sex could affect health been examined in detail. Either the satisfaction from partner). PSSW was not a significant predictor desire to engage in sexual activity (wanting) or the pleasure that of relative orgasm satisfaction. activity confers (liking) could contribute positively to health Study 3 showed that both the two subscales of the PSSLW status. We suggest that teasing apart how sexual wanting and scale predicted different real measures of sexual behavior. This sexual liking relate to health and well-being is an important next study showed that the scale had adequate predictive validity to step. assess sexual quality of specific relationships. Beyond contributing potentially useful measures for inclu- sion in research on sex and sexuality, the liking and wanting scales have potential applications to counseling. As well as being General Discussion of basic scientific value, providing knowledge of dispositional differences in wanting but not liking or vice versa may also be In this work, we developed and validated a short measure to useful to couples that are failing to effectively communicate distinguishbetweenPSSLandPSSWforbothmenandwomen. where their sexual differences lie. Identifying a sexual mismatch Study 1 showed that the PSSL and PSSW subscales had both (in liking or wanting or both) between members of a couple strong internal and test–retest reliability. Study 2 showed that could allow for more accurate counseling recommendations. As the measure had satisfactory construct validity. Lastly, Study 3 mentioned, high levels of sexual liking or sexual wanting, as well illustrated that the PSSLW scale had good predictive validity. as the degree to which these two measures are matched within We have suggested that these measures may be used to help couples, might have important physical, mental, and emotional explain low sexual frequency relative to reported sexual enjoy- health correlates. The PSSLW scale may help us to answer the ment within a relationship. More generally, our measures can be question of whether it is the general wanting of sex and the used to track changes in wanting and liking as a function of age carrying out of those desires that is beneficial within a relation- and relationship duration, and to examine these differences ship or whether it is the more subtle distinction of wanting and across genders. Our PSSLW scale may be used to identify pre- liking of sex with our current partner that has a positive impact on viously undiscovered predictors of relationship longevity and both individual well-being and the well-being of the couple as a success by answering such questions as,‘‘Do relationships with unit. passionate beginnings tend to maintain the for longer or exhibit more precipitous declines than those that begin in a more Acknowledgment The authors gratefully acknowledge the financial quiescent fashion?’’More work is required to examine how sex- support of a National Science Foundation Graduate Research Fellow- ship to Tamar Krishnamurti. ual liking and wanting change over time within relationships and whether gender differences in liking and wanting increase, decrease, or remain the same with aging, relationship duration, References and a number of other external factors. Having separate measures of both sexual liking and sexual Apt, C., & Hurlbert, D. F. (1992). Motherhood and female sexuality wanting can also help us to analyze the impact of situational beyond one year postpartum: A study of military . Journal of and Therapy, 18 determinants of both factors, beyond the impact of age and , 104–114. Baumeister, R. F. (2000). Gender differences in erotic plasticity: The relationship variables. Yet, levels of sexual liking and sexual female sex drive as socially flexible and responsive. Psychological wanting within the relationship may, in addition, be reflective of a Bulletin, 126, 375–389. more dispositional trait. We predict that baseline levels of sexual Baumeister, R. F., Catanese, K. R., & Vohs, K. (2001). Is there a gender liking and wanting in any relationship may be more reflective of difference in strength of sex drive? Theoretical views, conceptual distinctions, and a review of relevant evidence. Personality and an individual’s experiential capacity whereas changes in sexual Social Psychology Review, 5, 242–273. liking and wanting may be more susceptible to external influ- Bentler, P. M. (1990). Comparative fit indexes in structural models. ences. We saw gender differences in levels of sexual wanting, but Psychological Bulletin, 107, 238–246. Blanchflower,D.,&Oswald,A.(2004).Money,sex,andhappiness:An not in levels of sexual liking. This disconnect in the degree of empirical study. Scandinavian Journal of Economics, 106, 393–415. sexual wanting and sexual liking among women may help Brauer, L. H., Cramblett, M. J., Paxton, D. A., & Rose, J. E. (2001). explain some of the mixed results with respect to sex drive dif- Haloperidol reduces smoking of both nicotine-containing and ferences between genders (for a review of the mixed evidence, denicotinized cigarettes. Psychopharmacology, 159, 31–37. see Baumeister et al., 2001). Brauer, L., & de Wit, H. (1997). High dose pimozide does not block amphetamine-induced euphoria in normal volunteers. Pharma- In other research, sexual frequency has been shown to confer cology, Biochemistry and Behavior, 56, 265–272. indirect physical and emotional health benefits for both men and Breiter, H. C., Gollub, R. L., Weisskoff, R. M., Kennedy, D. N., Makris, women that are important to understand but little studied (Davey- N., Berke, J. D., et al. (1997). Acute effects of cocaine on human Neuron, 19 Smith, Frankel, & Yarnell, 1997; Laumann et al., 1994). It has not brain activity and . , 591–611. Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing been shown, for example, that the causality of the relationship model fit. In K. A. Bollen & J. S. Long (Eds.), Testing structural runs from frequency of sex to health; nor have the mechanisms by equation models (pp. 136–162). Newsbury Park, CA: Sage. 123 476 Arch Sex Behav (2012) 41:467–476

Call, V., Sprecher, S., & Schwartz, P. (1995). The incidence and frequency Kahneman, D., Krueger, A., Schkade, D., Schwarz, N., & Stone, A. of marital sex in a national sample. Journal of and the (2004). Toward national well-being accounts. American Economic , 57, 639–650. Review, 94, 429–434. Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant Krafft-Ebing, R. V. (1965). Psychopathia sexualis. New York: Putnam. validation by the multitrait–multimethod matrix. Psychological (Original work published 1886) Bulletin, 56, 81–105. Laan, E., & Everaerd, W. (1995). Determinants of female sexual arousal: Chivers, M. L., & Bailey, J. M. (2004). A sex difference in features that Psychophysiological theory and data. Annual Review of Sex Research, elicit genital response. Biological Psychology, 70, 115–120. 6, 32–76. Chivers, M. L., Rieger, G., Latty, E., & Bailey, J. M. (2004). A sex Laumann, E. O., Gagnon, J. H., Michael, R. T., & Michaels, S. (1994). difference in the specificity of arousal. Psychological Science, 15, The social organization of sexuality: Sexual practices in the United 736–744. States. Chicago: University of Chicago Press. Chivers, M. L., Seto, M. C., Lalumie`re, M. L., Laan, E., & Grimbos, T. Lucas, F., & Bellisle, F. (1987). The measurement of food preferences in (2010). Agreement of self-reported and genital measures of sexual humans: Do taste-and-spit test predict consumption. Physiology & arousal in men and women: A meta-analysis. Archives of Sexual Behavior, 39, 739–743. Behavior, 39, 5–56. Meston, C. M., Trapnell, P. D., & Gorzalka, B. B. (1998). Ethnic, gender, Cohen, L. L., & Shotland, R. L. (1996). Timing of first sexual intercourse and length-of-residency influences on sexual knowledge and atti- in a relationship: Expectations, experiences, and perceptions of tudes. Journal of Sex Research, 35, 176–188. others. Journal of Sex Research, 33, 291–299. Pecin˜a,S.,Berridge,K.C.,&Parker,L.A.(1997).Pimozidedoesnotshift Cox, D. N., Perry, L., Moore, P. B., Vallis, L., & Mela, D. J. (1999). Sensory palatability: Separation of anhedonia from sensorimotor suppression and hedonic associations with macronutrient and energy intakes of by taste reactivity. Pharmacology, Biochemistry and Behavior, 58, lean and obese consumers. International Journal of Obesity, 23, 801–811. 403–410. Rao, K. V., & DeMaris, A. (1995). Coital frequency among married and Davey-Smith, G., Frankel, S., & Yarnell, J. (1997). Sex and death: Are cohabiting couples in the U.S. Journal of Biosocial Science, 27, they related? Findings from the Caerphilly Cohort Study. British 135–150. Medical Journal, 315, 1641–1645. Smith, K. S., & Berridge, K. C. (2007). Opioid limbic circuit for reward: Elffers, H., Bethlehem, B., & Gill, R. (1978). Indeterminacy problems Interaction between hedonic hotspots of nucleus accumbens and and the interpretation of factor analysis results. Statistica Neer- ventral pallidum. Journal of Neuroscience, 27, 1594–1605. landica, 32, 181–199. Spector, I. P., Carey, M. P., & Steinberg, L. (1996). The sexual desire Finlayson, G., King, N., & Blundell, J. (2007). Liking vs. wanting food: inventory: Development, factor structure, and evidence of reli- Importance for human appetite control. Neuroscience and Biobe- ability. Journal of Sex and Marital Therapy, 22, 175–190. havioral Reviews, 31, 987–1002. Toates, F. (2009). An integrative theoretical framework for understand- Fisher, W. A., & Byrne, D. (1978). Individual differences in affective, ing sexual motivation, arousal, and behavior. Journal of Sex evaluative, and behavioral responses to . Journal of Applied Research, 46, 168–193. Social Psychology, 8, 355–365. Treit, D., & Berridge, K. C. (1990). A comparison of benzodiazepine, Greenblat, C. S. (1983). The salience of sexuality in the early years of dopamine, and serotonin agents in the taste-reactivity paradigm. marriage. Journal of Marriage and the Family, 45, 289–299. Pharmacology, Biochemistry and Behavior, 37, 451–456. Hendrick, S. (1988). A generic measure of relationship satisfaction. Tucker, L., & Lewis, C. (1973). The reliability coefficient for maximum Journal of Marriage and the Family, 50, 93–98. likelihood factor analysis. Psychometrika, 38, 1–10. Hobbs, M., Remington, B., & Glautier, S. (2005). Dissociation of Whitley, M. P., & Paulson, S. B. (1975). Assertiveness and sexual wanting and liking for alcohol in humans: A test of the incentive- satisfaction in employed professional women. Journal of Marriage sensitisation theory. Psychopharmacology, 178, 493–499. and the Family, 37, 573–581. Hudson, W. W., Harrison, D. F., & Crosscup, P. C. (1981). The index of Wyvell, C. L., & Berridge, K. C. (2000). Intra-accumbens amphetamine sexual satisfaction. Journal of Sex Research, 17, 157–174. increases the conditioned incentive salience of sucrose reward: Hunt, S. M. (1974). Sex differences: A biological and psychological Enhancement of reward ‘‘wanting’’ without enhanced ‘‘liking’’ or analysis of the differences between men and women.RoyalLeam- response reinforcement. Journal of Neuroscience, 20, 8122–8130. ington Spa, UK: Vernon Scott Associates Ltd.

123