The organization of sexual preferences

Kodi B. Arfer and Nicholas R. Eaton
Created 7 Nov 2013 • Last modified 12 Mar 2019

N.B. This manuscript is a work in progress.

Introduction

What kinds of sexual activities and sexual partners are people attracted to? As in all domains of human life, such as food, movies, and academic disciplines, there are widespread individual differences in what people like. The diversity of people's "sexual preferences"—by which we mean preferences concerning sexual activity, whether for particular kinds of partners, particular sexual acts, or even moods or themes, such as vulnerability or purity—is perhaps most obvious from surveys of Internet pornography (e.g., Ogas & Gaddam, 2011) and Internet communities devoted to particular sexual interests (e.g., Scorolli, Ghirlanda, Enquist, Zattoni, & Jannini, 2007) or even the lack of sexual interests (Chasin, 2011). It seems reasonable to expect that preferences will be interrelated; for example, people who are interested in dominance or submission might also be interested in sexual scenes involving bondage or leather. How, then, are sexual preferences organized? In other words, how are sexual preferences related to each other?

In research and popular discourse, sexual preferences are typically organized around sexual orientations, which, in turn, are typically construed in terms of partner gender and discrete categories. Debate continues on exactly which sexual orientations exist (Rosenthal, Sylva, Safron, & Bailey, 2012; Savin-Williams & Vrangalova, 2013; Kuper, Nussbaum, & Mustanski, 2012; Bogaert, 2006; Seto, 2012). Other controversy includes whether discrete orientation categories should be replaced with continuous scales (Ellis, Burke, & Ames, 1987; Haslam, 1997) and whether continuous variation exists on top of latent discrete categories (Gangestad, Bailey, & Martin, 2000).

While there is debate over the details of how to conceptualize sexual orientation, we see two fundamental issues that have rarely been addressed. First, focusing on sexual orientation means focusing on the features of a prospective sexual partner, and usually only one feature, namely, the partner's gender. One does not have the opportunity to consider preferences for less concrete partner features, such as personality or social status, nor different sexual activities, such as masturbation or anal sex, nor moods or themes, such as BDSM. People with the same sexual orientation need not agree on any of these. Second, the construct of sexual orientation has not been developed in an integrative empirical fashion with the idea of finding structure underlying the totality of people's preferences. By contrast, consider research into personality, intelligence, and psychopathology. Personality researchers have found that hundreds or thousands of descriptive terms can be reduced to a few personality traits (Goldberg, 1990). Intelligence researchers have found a single dimension of fluid intelligence that can explain performance on a wide variety of tasks (Gottfredson, 1997). Psychopathology researchers have found that the many psychiatric symptoms people experience can be summarized with a few dimensions, such as internalizing, externalizing, and thought disorder (Kotov et al., 2017). Perhaps the whole universe of sexual preferences, too, can be characterized with such a coherent underlying structure.

In this study, we sought to bring a personality-inspired and statistically rich approach to the study of sexual preferences. We used a self-report form to survey preferences for a wide variety of characteristics of sexual acts and partners. We analyzed these items with two kinds of unsupervised statistical learning methods to search for underlying structure and show how different preferences are related to each other.

Method

See http://arfer.net/projects/galaxy for task code (including the complete instrument) and data-analysis code.

Subjects

We recruited and ran subjects on Amazon Mechanical Turk (MTurk), a data-collection website, in April, May, and July of 2014. MTurk "workers" are ordinary Internet users aged 18 and above who volunteer to complete short unskilled online tasks for nominal fees. Increasingly, MTurk is being used by researchers to study human behavior. For example, Suri and Watts (2011) studied social networks, and Eriksson and Simpson (2010) examined gender differences in risk preferences. Evidence that MTurk can provide data comparable in quality to that produced in conventional laboratory settings is provided by studies such as Buhrmester, Kwang, and Gosling (2011), which replicated previous estimates of the retest reliability of several personality measures. While samples of Americans on MTurk are not random samples of the US population, they are likely to be more diverse (Ipeirotis, 2010), and therefore more representative of the US population, than the undergraduate subject pools used in most studies with human subjects. Moreover, we expected that subjects using an anonymous online platform would be more honest about their sexual preferences and less influenced by social desirability.

A total of 1,001 MTurk users completed the study. The study was described as "a survey about sexual feelings and sexual behavior". Subjects were required to live in the United States. The median completion time was 13 minutes. Subjects were compensated with $2.

Instrument

The items analyzed in this study were part of a larger questionnaire. Subjects were told "We would like you to consider how sexually appealing you find each of these activities." They were presented with a number of short descriptions and rated each on a 7-point scale with anchors "Not at all appealing" (1), "Somewhat appealing" (4), and "Very appealing" (7). There were 94 items in total, 9 of which were presented only to subjects who described their gender as "Male" and 5 of those who selected "Female". See the appendix for complete instructions and item text. The items were presented in a random order for each subject. In line with our focus on sexual preferences, instructions emphasized that subjects should "answer using only your sexual and romantic feelings", as opposed to considering whether they would actually be willing to perform a given activity (e.g., one cannot consent to being raped, by definition, but one could still enjoy a fantasy of such). Here is an outline of the various items.

  • Items 1–20 concern specific sexual acts such as masturbation, coitus, anal copulation, and oral sex.
  • Items 21–27 concern semen specifically.
  • Items 28–45 concern fetishes such as bondage, leather, and urine play. Also included are items about rape, both rape in general and sex with an unconscious person in particular.
  • Items 46–50 concern pornography.
  • Items 51–61 concern the relationship with a sex partner and feelings towards them.
  • Items 62–69 concern the size of partner body features such as the penis, breasts, and feet.
  • Items 70–80 concern the gender, gender typicality, and age of a partner.
  • Items 81–94 concern other partner features such as personality and sexual experience. The final ten items are based on the short Big Five personality test of Gosling, Rentfrow, and Swann (2003).

At the end of the questionnaire, two questions related to sexual preferences were repeated as validity items: one about fellating someone else and one about sex with a friend. A third validity item asked "How honest were you able to be while answering this survey? (We won't withhold payment based on your answer to this question.)" Subjects responded on a 5-point scale with anchors "Not at all honest", "Somewhat honest", and "Entirely honest".

Demographic questions (including items about gender, age, etc.) were asked at the beginning of the questionnaire. The gender item asked "What is your gender?" and had four fixed options—"Male", "Female", "Transgender (male-to-female)", and "Transgender (female-to-male)"—and a free-response option labeled "Other".

Analytic sample

To reduce the effect of careless responses on our results, we excluded subjects from analysis on the basis of response time and their responses to the three validity items in part 8. Subjects were excluded if they completed the task in less than 7 minutes (n = 46), they rated their honesty as "Somewhat honest" or less (n = 20), or their responses to both repeated items differed from their original responses by 3 units or more (n = 5). One subject was excluded because she said she had selected the wrong gender, meaning she had been asked the wrong preferences questions. Finally, since this study analyzes men and women separately, but without trying to distinguish the effects of anatomy, endocrinology, social role, gender identity, and so on, we excluded all subjects who chose an option for the gender item other than "Male" or "Female" (n = 15). The resulting analytic sample comprised 923 subjects.

Data analysis

We use two basic methods—clustering and factor analysis, similarly to Forbes et al. (2017)—to examine how the preference items are related to each other, and in particular, how they can be grouped. We try each approach separately within each sex.

The clustering approach is theoretically simple and requires no modeling or scale assumptions beyond the assumption of an ordinal scale for each item and independent sampling of subjects. Clustering is usually used to group cases (subjects), but we use it to group variables (items). We use agglomerative mean-linkage hierarchical clustering (function hclust in the R package stats) with the distance metric d(X, Y) = 1 − |k(X, Y)|, where k(X, Y) is the Kendall correlation of X with Y. We use Kendall correlation, a rank-based measure, to avoid having to assign meaning to the distance between points on each item's 7-point rating scale (to account for ties, R uses the τB metric of Kendall, 1945, which removes tied pairs from the quotient used to calculate τ). Mean-linkage hierarchical clustering proceeds iteratively by first assigning every object to its own cluster, so there are q clusters for q objects. Then, for every pair of clusters A and B, the mean of the pairwise distances between objects in A and B is computed, and the two clusters with the least such mean are combined into one cluster, leaving q − 1 clusters. Clusters continue to be combined two at a time until all q objects are in a single cluster. Given our distance metric, each of our clusters is characterized by a mean absolute Kendall correlation. We present the results backwards, starting from the case of 1 cluster and describing the results for 2, 3, 4, 5, 6, and 7 clusters. The effect is that for each k, the analysis with k clusters has all the same clusters as the analysis with k − 1 clusters, except that one of the old clusters has had items removed from it to form the new cluster.

Exploratory factor analysis is a method that is more familiar to social scientists but relies on concrete distributional assumptions about the data. We take a bass-ackwards approach (Goldberg, 2006), in which we fit a 2-factor solution, followed by a 3-factor solution, and so on up to 7 factors, with each solution as a separate model. We then save factor scores and compare them between levels. Each factor analysis (function fa in the R package psych) is fit with maximum likelihood estimation on a polychoric correlation matrix and uses orthogonal varimax rotation.

Clustering and factor analyses are not usually methods that are used to address the same problems. But by clustering variables, rather than clustering subjects, we make clustering, like factor analysis, an unsupervised way to see how variables can be grouped on the basis of their interrelations. The advantage of using both methods is that clustering is simple whereas factor analysis is complex. The interpretation of clusters does not depend on modeling assumptions such as the shape of items' distributions, but can only speak to relationships between items. Factor analysis relies on the strong assumption of the common-factor model, but it makes stronger claims, including a loading of each item on each factor, and a score of each subject on each factor.

Results

Demographics

Of the analytic sample, 459 subjects (50%) were female. Ages ranged from 18 to 82 (median 31). Subjects were 80% white, 10% black, 8% Asian, 6% Hispanic, and 2% Native American (they were allowed to select more than one of these categories). With regard to education, 12% had a graduate degree, 39% had an undergraduate degree, 37% had had some college education, 11% had graduated from high school, and the remaining 1% had not completed high school. The median annual-income category was $20,000 to $29,999. Most subjects (97%) identified as native speakers of English, and the remainder as fluent non-native speakers of English.

Men's preferences

The cluster analysis for men is shown in Table 1. See Figure 1 for a dendrogram. The initial cluster of all items could be characterized as interest in men, going by the items with the greatest mean absolute correlation. Of the other clusters, all but cluster 6 form by removing items directly from cluster 1. Cluster 2 does not have an obvious interpretation—it seems to characterize interest in a partner with personality traits that are typically undesirable. It may be related to subjects' lack of concern for partner personality. Cluster 3 concerns interest in women. Cluster 4 comprises mildly socially proscribed items (kink) such as voyeurism and bondage. Cluster 5 comprises strongly proscribed items, where consent is lacking or expected to be lacking, including rape, incest, and sex with children (including teenagers). Cluster 6 is a single-item cluster, containing the item about sex with a virgin, which is taken from cluster 4. Finally, cluster 7 contains two items: sex with a masculine woman and sex with a 70-year-old. In order, the seven clusters can be described as: men, unpersonability, women, kink, nonconsent, virginity, and unusual partners.

Table 1. Clusters of men's sexual preferences. Shown are the top five items of each cluster (columns) in each stage of clustering (row groups), scored by mean absolute Kendall correlation with all other items in the cluster. The top of each cluster shows the total number of items in the cluster (q) as well as the mean and SD of all pairwise absolute Kendall correlations. For every cluster in this table, the sign of the correlation of the top item with the lower four items is positive.
1 (Men) 2 (Unpersonability) 3 (Women) 4 (Kink) 5 (Nonconsent) 6 (Virginity) 7 (?)
q 89, M .14, SD .13            
.21 Man            
.21 Penis in man's anus            
.21 Fellating            
.20 Swallowing semen            
.20 Masturbating a man            
q 80, M .15, SD .13 q 9, M .24, SD .09          
.23 Man .28 Non-open partner          
.23 Penis in man's anus .26 Unconscientious partner          
.22 Fellating .24 Small buttocks          
.22 Swallowing semen .24 Neurotic partner          
.22 Masturbating a man .24 Introverted partner          
q 56, M .19, SD .16 q 9, M .24, SD .09 q 24, M .22, SD .11        
.28 Man .28 Non-open partner .29 Masturbating a woman        
.28 Penis in man's anus .26 Unconscientious partner .28 Coitus        
.28 Fellating .24 Small buttocks .27 Giving pleasure        
.28 Swallowing semen .24 Neurotic partner .27 Feminine woman        
.27 Masturbating a man .24 Introverted partner .25 Sex in relationship        
q 25, M .35, SD .24 q 9, M .24, SD .09 q 24, M .22, SD .11 q 31, M .21, SD .10      
.50 Man .28 Non-open partner .29 Masturbating a woman .26 Voyeurism in secret      
.49 Receiving semen in mouth .26 Unconscientious partner .28 Coitus .26 Insulting      
.48 Swallowing semen .24 Small buttocks .27 Giving pleasure .25 Voyeurism in open      
.48 Large penis .24 Neurotic partner .27 Feminine woman .25 Tying up      
.48 Fellating .24 Introverted partner .25 Sex in relationship .25 Ejaculating into anus      
q 19, M .50, SD .21 q 9, M .24, SD .09 q 24, M .22, SD .11 q 31, M .21, SD .10 q 6, M .30, SD .09    
.62 Man .28 Non-open partner .29 Masturbating a woman .26 Voyeurism in secret .36 Raping    
.60 Receiving semen in mouth .26 Unconscientious partner .28 Coitus .26 Insulting .34 Unconscious partner    
.60 Large penis .24 Small buttocks .27 Giving pleasure .25 Voyeurism in open .30 14-year-old    
.60 Fellating .24 Neurotic partner .27 Feminine woman .25 Tying up .28 Being raped    
.59 Swallowing semen .24 Introverted partner .25 Sex in relationship .25 Ejaculating into anus .27 Incest    
q 19, M .50, SD .21 q 9, M .24, SD .09 q 24, M .22, SD .11 q 30, M .22, SD .10 q 6, M .30, SD .09 q 1, M NaN, SD NaN  
.62 Man .28 Non-open partner .29 Masturbating a woman .26 Voyeurism in secret .36 Raping Virgin  
.60 Receiving semen in mouth .26 Unconscientious partner .28 Coitus .26 Insulting .34 Unconscious partner    
.60 Large penis .24 Small buttocks .27 Giving pleasure .26 Voyeurism in open .30 14-year-old    
.60 Fellating .24 Neurotic partner .27 Feminine woman .26 Tying up .28 Being raped    
.59 Swallowing semen .24 Introverted partner .25 Sex in relationship .25 Ejaculating into anus .27 Incest    
q 17, M .58, SD .13 q 9, M .24, SD .09 q 24, M .22, SD .11 q 30, M .22, SD .10 q 6, M .30, SD .09 q 1, M NaN, SD NaN q 2, M .25, SD .00
.67 Man .28 Non-open partner .29 Masturbating a woman .26 Voyeurism in secret .36 Raping Virgin .25 Masculine woman
.66 Receiving semen in mouth .26 Unconscientious partner .28 Coitus .26 Insulting .34 Unconscious partner   .25 70-year-old
.65 Large penis .24 Small buttocks .27 Giving pleasure .26 Voyeurism in open .30 14-year-old    
.65 Swallowing semen .24 Neurotic partner .27 Feminine woman .26 Tying up .28 Being raped    
.65 Fellating .24 Introverted partner .25 Sex in relationship .25 Ejaculating into anus .27 Incest    
hier-tree-m.png
Figure 1. Dendrogram of the cluster analysis of men's sexual preferences. Height is defined such that among all the items grouped together below a given height h, every pair has an absolute Kendall correlation of at least 1 - h.

For factor analysis, we first count zero cells because of our use of polychoric correlation. Overall, 22% of all pairs of responses choices never occur; 96% of item pairs have at least one of these zeros in their frequency table. The factor analyses are depicted in Table 2 and Table 3. At level 1, when there is only one factor, the lone factor is driven by interest in men, particularly fellating another man. This factor reappears consistently in the subsequent analyses. At level 2, factor 2 is associated with interest in women, pornography, and insertive oral sex. At level 3, factor 2 resembles a kink factor more than a women factor, while the new factor 3 is related to tenderness, headed by a negative loading for sex with an 8-year-old. Level 4 restores the factor 2's focus on women and leaves factor 4 as a kink factor or a more general willingness to explore sexually. At level 5, factor 5 recalls the "undesirable personality traits" cluster from the cluster analysis. At level 6, factor 6 appears to concern nonmonogamous sex; its factor scores are highly correlated with those of the previous level's factor 5, and resembles the pornography cluster from the cluster analysis. Finally, factor 7 concerns the anus, especially women's anuses. In order, the seven factors can be described as: men, women, tenderness, kink, undesirable personality traits, nonmonogamy, and the female anus.

Table 2. Factor analyses of men's sexual preferences. Shown are the top five items of each factor (columns) in each factor analysis (row groups), scored by factor loading (which is sorted by absolute value).
1 (Men) 2 (Women) 3 (Tenderness) 4 (Kink) 5 (Unpersonability) 6 (Nonmonogamy) 7 (Female Anus)
           
.96 Man            
.95 Receiving semen in mouth            
.94 Large penis            
.94 Fellating            
.93 Swallowing semen            
         
.96 Man .67 Ejaculating into mouth          
.94 Receiving semen in mouth .65 Porn video, male-female          
.94 Fellating .64 Masturbating a woman          
.94 Large penis .62 Penis in woman's anus          
.93 Swallowing semen .62 Porn video, female-female          
       
.97 Man .70 Penis in woman's anus -.77 8-year-old        
.95 Fellating .69 Ejaculating into anus .69 Giving pleasure        
.95 Large penis .64 Voyeurism in secret .69 Sex in relationship        
.94 Masculine man .64 Insulting .66 Being sexually desired        
.94 Receiving semen in mouth .63 Acquaintance .66 Feeling in love        
     
.95 Man .66 Giving cunnilingus -.77 8-year-old .73 Acquaintance      
.94 Receiving semen in mouth .61 Masturbating a woman .67 Sex in relationship .68 Stranger      
.94 Fellating .57 Coitus .66 Giving pleasure .68 Friend      
.92 Large penis .57 Ejaculating into vagina .66 Feeling in love .65 Controlling      
.92 Swallowing semen .57 Giving a woman anilingus .64 Being sexually desired .63 Insulting      
   
.95 Man .64 Giving cunnilingus -.78 8-year-old .65 Ejaculating into anus .61 Non-open partner    
.94 Receiving semen in mouth .57 Masturbating a woman .67 Giving pleasure .65 Controlling .60 Unconscientious partner    
.94 Fellating .54 Giving a woman anilingus .67 Sex in relationship .64 Insulting .57 Neurotic partner    
.92 Large penis .54 Coitus .66 Feeling in love .64 Tying up .53 Introverted partner    
.91 Swallowing semen .53 Woman .65 Being sexually desired .63 Ejaculating into mouth .50 Disagreeable partner    
 
.95 Man .63 Giving cunnilingus .72 Feeling in love .63 Receiving pain .62 Non-open partner .70 Acquaintance  
.93 Receiving semen in mouth .58 Masturbating a woman .69 Sex in relationship .63 Tying up .58 Unconscientious partner .70 Ejaculating into mouth  
.93 Fellating .58 Woman .68 Emotional closeness .62 Being tied up .57 Neurotic partner .69 Stranger  
.92 Large penis .57 Coitus .65 Kissing .57 Causing pain .52 Introverted partner .68 Group sex  
.91 Swallowing semen .56 Ejaculating into vagina .64 Giving pleasure .55 Controlling .50 Disagreeable partner .68 Friend  
.96 Man .54 Giving cunnilingus .73 Feeling in love .73 Being tied up .64 Non-open partner .69 Acquaintance .82 Ejaculating into anus
.94 Fellating .53 Woman .70 Sex in relationship .68 Receiving pain .58 Unconscientious partner .68 Friend .73 Penis in woman's anus
.94 Receiving semen in mouth .52 Masturbating a woman .68 Emotional closeness .64 Being controlled .56 Neurotic partner .68 Group sex .59 Giving a woman anilingus
.93 Large penis .51 Coitus .65 Giving pleasure .61 Tying up .55 Introverted partner .67 Ejaculating into mouth .44 Receiving anilingus
.92 Masturbating a man .47 Feminine woman .64 Kissing .59 Being insulted .53 Large feet .67 Stranger .33 Ejaculating into mouth
Table 3. Correlation of factor scores for men's sexual preferences, between factors at each level of the bass-ackwards procedure (rows) with those at the previous level (columns). Off-diagonal elements with absolute value of 1/3 or more are in bold.
  1 2 3 4 5 6
           
1 .99          
2 -.29          
         
1 .98 -.24        
2 .12 .83        
3 -.53 .61        
       
1 .99 .07 -.39      
2 -.12 .56 -.00      
3 -.38 -.00 .99      
4 -.06 .86 .27      
     
1 .99 .02 -.39 -.12    
2 .08 .99 -.08 -.02    
3 -.40 -.02 .99 .23    
4 -.05 .17 .18 .94    
5 .07 .03 -.06 .36    
   
1 .99 .06 -.39 -.09 .07  
2 .01 .98 .03 .09 .01  
3 -.35 -.04 .95 .17 -.08  
4 .26 .23 -.48 .47 -.06  
5 .13 .05 -.20 -.02 .97  
6 -.27 -.08 .56 .82 .11  
 
1 .99 -.02 -.36 .24 .13 -.25
2 .10 .91 -.02 -.01 .15 -.05
3 -.36 .03 .99 -.27 -.16 .40
4 .19 .00 -.19 .93 .15 -.13
5 .10 -.00 -.16 .15 .96 -.04
6 -.23 .05 .38 -.11 -.05 .99
7 .03 .48 -.11 .39 -.20 .11

Women's preferences

The cluster analysis for women is shown in Table 4. See Figure 2 for a dendrogram. The initial cluster is dominated by interest in women, similarly to how the initial cluster for men was dominated by interest in men. Cluster 2 concerns tenderness, such as sex in a committed relationship and intimacy; oddly, being masturbated by a partner appears here. Cluster 3 takes the now-familiar unpersonability group out of cluster 1. Cluster 4 concerns nonconsent, and is also formed from cluster 1. Cluster 5 takes a single item, sex with a 70-year-old person, out of cluster 3. Cluster 6 is the first appearance of a cluster for interest in men; it draws from cluster 1. Finally, cluster 7 removes body characteristics from cluster 3, including feet, small buttocks, and small penises. In order, the seven clusters can be described as: women, tenderness, nonconsent, senior partners, men, and small body parts.

Table 4. Clusters of women's sexual preferences. Shown are the top five items of each cluster (columns) in each stage of clustering (row groups), scored by mean absolute Kendall correlation with all other items in the cluster. The top of each cluster shows the total number of items in the cluster (q) as well as the mean and SD of all pairwise absolute Kendall correlations. For every cluster in this table, the sign of the correlation of the top item with the lower four items is positive.
1 (Women) 2 (Tenderness) 3 (Unpersonability) 4 (Nonconsent) 5 (Seniors) 6 (Men) 7 (Smallness)
q 85, M .15, SD .11            
.22 Giving cunnilingus            
.22 Group sex            
.21 Woman            
.21 Porn videos or pictures            
.21 Porn video, female-female            
q 73, M .17, SD .11 q 12, M .26, SD .10          
.25 Giving cunnilingus .31 Sex in relationship          
.24 Group sex .30 Emotional closeness          
.23 Porn video, female-female .29 Feeling in love          
.23 Porn videos or pictures .27 Giving pleasure          
.23 Masturbating a woman .27 Being sexually desired          
q 61, M .20, SD .12 q 12, M .26, SD .10 q 12, M .18, SD .08        
.27 Giving cunnilingus .31 Sex in relationship .22 Neurotic partner        
.27 Group sex .30 Emotional closeness .22 Small feet        
.26 Porn video, female-female .29 Feeling in love .21 Small buttocks        
.26 Woman .27 Giving pleasure .21 Non-open partner        
.26 Masturbating a woman .27 Being sexually desired .21 Unconscientious partner        
q 55, M .21, SD .12 q 12, M .26, SD .10 q 12, M .18, SD .08 q 6, M .27, SD .11      
.29 Giving cunnilingus .31 Sex in relationship .22 Neurotic partner .31 14-year-old      
.28 Group sex .30 Emotional closeness .22 Small feet .30 Raping      
.28 Porn video, female-female .29 Feeling in love .21 Small buttocks .27 Incest      
.28 Woman .27 Giving pleasure .21 Non-open partner .27 Being raped      
.27 Masturbating a woman .27 Being sexually desired .21 Unconscientious partner .23 8-year-old      
q 55, M .21, SD .12 q 12, M .26, SD .10 q 11, M .20, SD .08 q 6, M .27, SD .11 q 1, M NaN, SD NaN    
.29 Giving cunnilingus .31 Sex in relationship .23 Small feet .31 14-year-old 70-year-old    
.28 Group sex .30 Emotional closeness .23 Neurotic partner .30 Raping      
.28 Porn video, female-female .29 Feeling in love .22 Small buttocks .27 Incest      
.28 Woman .27 Giving pleasure .22 Non-open partner .27 Being raped      
.27 Masturbating a woman .27 Being sexually desired .21 Unconscientious partner .23 8-year-old      
q 42, M .25, SD .12 q 12, M .26, SD .10 q 11, M .20, SD .08 q 6, M .27, SD .11 q 1, M NaN, SD NaN q 13, M .26, SD .13  
.35 Giving cunnilingus .31 Sex in relationship .23 Small feet .31 14-year-old 70-year-old .33 Fellating  
.33 Woman .30 Emotional closeness .23 Neurotic partner .30 Raping   .30 Receiving semen in mouth  
.33 Masturbating a woman .29 Feeling in love .22 Small buttocks .27 Incest   .29 Swallowing semen  
.33 Porn video, female-female .27 Giving pleasure .22 Non-open partner .27 Being raped   .29 Large penis  
.31 Group sex .27 Being sexually desired .21 Unconscientious partner .23 8-year-old   .29 Coitus  
q 42, M .25, SD .12 q 12, M .26, SD .10 q 5, M .25, SD .08 q 6, M .27, SD .11 q 1, M NaN, SD NaN q 13, M .26, SD .13 q 6, M .25, SD .08
.35 Giving cunnilingus .31 Sex in relationship .28 Neurotic partner .31 14-year-old 70-year-old .33 Fellating .30 Small feet
.33 Woman .30 Emotional closeness .28 Unconscientious partner .30 Raping   .30 Receiving semen in mouth .27 Large feet
.33 Masturbating a woman .29 Feeling in love .26 Non-open partner .27 Incest   .29 Swallowing semen .27 Small buttocks
.33 Porn video, female-female .27 Giving pleasure .25 Disagreeable partner .27 Being raped   .29 Large penis .24 Feet
.31 Group sex .27 Being sexually desired .18 Introverted partner .23 8-year-old   .29 Coitus .23 Small penis
hier-tree-f.png
Figure 2. Dendrogram of the cluster analysis of women's sexual preferences. Height is defined such that among all the items grouped together below a given height h, every pair has an absolute Kendall correlation of at least 1 - h.

For factor analysis, we again count zero cells. Overall, 21% of all pairs of responses choices never occur; 94% of item pairs have at least one of these zeros in their frequency table. The factor analyses are depicted in Table 5 and Table 6. The first factor is driven by interest in women. At level 2, factor 2 looks like a tenderness factor, and also has items concerning interest in men. At level 3, factor 2 loses weight on items for interest in men and gains negative weight for nonconsent items such as incest and rape. The new factor 3 concerns interest in men, particularly semen, but it is positively related to the previous factor 1 (women). At level 4, factor 4 comprises interest in more feminine or less masculine men, such as female-to-male transgender people and men with small penises. It is negatively related to the previous factor 2 (tenderness), but positively with factor 3 (semen). At level 5, factor 5 concerns pornographic tropes such as sex with a stranger or a "bad boy", and also weights items for actually viewing pornography. It is strongly positively correlated with the previous factor 3 (semen). At level 6, factor 6 concerns the anuses, one's own and those of others. The final factor concerns domination and submission, but especially submission. In order, the seven factors could be described as: women, tenderness, semen, feminine men, pornography, the anus, and power.

Table 5. Factor analyses of women's sexual preferences. Shown are the top five items of each factor (columns) in each factor analysis (row groups), scored by factor loading (which is sorted by absolute value).
1 (Women) 2 (Tenderness) 3 (Semen) 4 (Feminine Men) 5 (Pornography) 6 (Anus) 7 (Power)
           
.81 Giving cunnilingus            
.78 Woman            
.77 Masturbating a woman            
.77 Porn video, female-female            
.77 Feminine woman            
         
.78 Giving cunnilingus .85 Being sexually desired          
.76 Porn video, female-female .74 Man          
.74 Group sex .73 Sex in relationship          
.74 Woman .72 Emotional closeness          
.74 Masturbating a woman .72 Coitus          
       
.94 Woman .84 Being sexually desired .70 Receiving semen in mouth        
.93 Masturbating a woman -.78 Incest .70 Swallowing semen        
.91 Feminine woman .77 Feeling in love .69 Fellating        
.89 Giving cunnilingus .75 Sex in relationship .61 Large penis        
.83 Large breasts .75 Emotional closeness .61 Receiving semen in anus        
     
.94 Woman .79 Emotional closeness .68 Receiving semen in mouth .57 Small penis      
.92 Masturbating a woman .79 Sex in relationship .68 Fellating .55 Feminine man      
.90 Feminine woman .78 Feeling in love .67 Swallowing semen .53 Small feet      
.87 Giving cunnilingus .78 Being sexually desired .66 Large penis .50 FtM      
.81 Large breasts .72 Giving pleasure .63 Being controlled .48 Large feet      
   
.94 Woman .79 Sex in relationship .89 Receiving semen in mouth .56 Feminine man .67 Stranger    
.92 Masturbating a woman .79 Emotional closeness .88 Swallowing semen .56 Small penis .64 Rebel    
.90 Feminine woman .78 Being sexually desired .60 Fellating .55 Small feet .63 Large penis    
.86 Giving cunnilingus .78 Feeling in love .57 Receiving semen in anus .49 Large feet .63 Acquaintance    
.81 Large breasts .73 Giving pleasure .47 Giving a man anilingus .48 FtM .61 Group sex    
 
.94 Woman .80 Emotional closeness .85 Receiving semen in mouth .62 Feminine man .67 Stranger .71 Receiving anilingus  
.92 Masturbating a woman .79 Sex in relationship .85 Swallowing semen .59 Small penis .64 Rebel .68 Giving a woman anilingus  
.90 Feminine woman .78 Being sexually desired .56 Fellating .52 FtM .63 Large penis .63 Giving a man anilingus  
.85 Giving cunnilingus .78 Feeling in love .46 Receiving semen in anus .52 Small feet .62 Acquaintance .46 Receiving penis in anus  
.81 Large breasts .73 Giving pleasure .40 Receiving semen in vagina .49 Neurotic partner .62 Group sex .41 Receiving semen in anus  
.94 Woman .81 Sex in relationship .85 Receiving semen in mouth .62 Feminine man .62 Porn video, male-female .69 Giving a woman anilingus .70 Being tied up
.92 Masturbating a woman .80 Emotional closeness .84 Swallowing semen .60 Small penis .61 Large penis .67 Receiving anilingus .70 Being controlled
.91 Feminine woman .79 Being sexually desired .56 Fellating .53 FtM .60 Masculine man .62 Giving a man anilingus .61 Receiving pain
.86 Giving cunnilingus .78 Feeling in love .45 Receiving semen in anus .52 Small feet .59 Stranger .41 Receiving penis in anus .58 Being insulted
.82 Large breasts .74 Giving pleasure .41 Receiving semen in vagina .49 Large feet .57 Acquaintance .35 Urinating on partner .56 Tying up
Table 6. Correlation of factor scores for women's sexual preferences, between factors at each level of the bass-ackwards procedure (rows) with those at the previous level (columns). Off-diagonal elements with absolute value of 1/3 or more are in bold.
  1 2 3 4 5 6
           
1 .99          
2 -.38          
         
1 .74 -.12        
2 -.38 .98        
3 .68 .03        
       
1 .99 -.05 -.03      
2 -.07 .98 -.17      
3 .12 .06 .93      
4 .10 -.67 .40      
     
1 .99 -.09 .04 .08    
2 -.06 .99 .01 -.50    
3 -.05 -.25 .60 .27    
4 .04 -.46 .02 .98    
5 .10 .16 .81 -.13    
   
1 .99 -.06 .01 .03 .05  
2 -.08 .99 -.23 -.46 .15  
3 -.00 -.12 .97 .00 -.03  
4 .07 -.44 .15 .97 -.05  
5 .05 .14 .08 -.07 .99  
6 .12 -.32 .31 .44 .01  
 
1 .99 -.04 -.00 .04 .09 .04
2 -.07 .99 -.13 -.44 .15 -.27
3 .00 -.14 .99 .07 .04 .08
4 .06 -.46 .08 .99 -.07 .25
5 -.00 .27 -.12 -.11 .88 -.10
6 .03 -.22 -.01 .19 -.01 .97
7 .02 -.24 .29 .11 .45 .26

Discussion

Our Internet sample of 459 men and 464 women exhibits a number of ways to group sexual preferences—some familiar, some novel—which vary between genders as well as between the two methods, clustering and factor analysis. A few groups reoccur frequently between methods and genders. Gender preferences—preferences for sex with men or women in particular—figure largely in every case, providing support for the traditional emphasis on gender preferences in the study of sexual preferences. Other reoccurring groups include tenderness, an interest in things such as sex in romantic relationship, giving one's partner sexual pleasure, and emotional closeness, and unpersonability, an interest in partners who have conventionally undesirable personality traits, such as low openness and high neuroticism.

In men, both methods show groups for preferences concerning men, women, kink, and unpersonability. The cluster analysis, but not the factor analysis, has a separate group for rape and presumed rape such as sex with an 8-year-old; the factor analysis puts these items at negative loadings in the tenderness factor. Possibly the tenderness factor is more of a social-desirability factor.

In women, both methods show groups for preferences concerning women, tenderness, and men, although the men group that appears in the factor analysis is dominated by feminine traits. The prominence of the tenderness group supports theories that intimacy and romance are particularly important for women (Meana, 2010). Meanwhile, a group for power relations that appears only for factor analysis and only at stage 7 provides weak support for theories that construe sexual interests in terms of power.

Notice that the group for interest in the opposite sex, separate from same-sex interest, appears more readily in men than women. This suggests that heterosexual and homosexual interest is more independent in men than women, which is reminiscent of findings that lesbians are more likely than gay men to report a different orientation over time (Ott, Corliss, Wypij, Rosario, & Austin, 2011; Mock & Eibach, 2012; Savin-Williams, Joyner, & Rieger, 2012).

There are several limitations inherent in the design of this study. Most obvious is that our results are shaped by the design of the questionnaire; while we attempted to be theoretically neutral, or at least to cover many different theories, there can be no truly neutral questionnaire. There are also the usual honesty and accuracy concerns with self-report, particularly when asking subjects about such sensitive matters as their interest in sex with an 8-year-old. Our use of an anonymous Internet sample, instead of a traditional subject pool, may have made subjects somewhat more willing to be honest. Phallometry could be a useful replacement or addition to self-report in men, but genital arousal does not exhaust the concept of sexual arousal (Toates, 2009), particularly in women (Chivers, Seto, Lalumière, Laan, & Grimbos, 2010), let alone sexual preference. Finally, more detailed measurement of gender identity, along with targeted sampling, would be necessary to examine sexual preferences among transgender people.

A more subtle issue is that sexual preferences may not be stable, especially in their details. As previously mentioned, people's reported orientation may change over the years. Preferences may also be influenced by circumstantial factors such as menstrual-cycle phase (Gildersleeve, Haselton, & Fales, 2014). Ariely, Loewenstein, and Prelec (2006) challenge the very idea of preferences by showing that whether people regard an experience of hearing poetry as work or play—and hence worth paying for or being paid for—can be influenced by prompting them to think of it as one or the other. Finding people's "enduring patterns" of sexual interests may require measuring them repeatedly in a variety of situations.

Appendix: preference items

  1. mast_solo_m [M] Masturbating alone (touching your penis; jacking off)
  2. mast_solo_f [F] Masturbating alone (touching your clitoris or vagina)
  3. coitus_m [M] Having vaginal sex with a woman (penetrating a woman with your penis)
  4. coitus_f [F] Having vaginal sex with a man (being penetrated by a man's penis)
  5. analpen_to_f [M] Giving anal sex to a woman (penetrating a woman's anus with your penis; topping)
  6. analpen_to_m [M] Giving anal sex to a man (penetrating a man's anus with your penis; topping)
  7. analpen_rec Receiving anal sex (being anally penetrated by a man's penis; bottoming)
  8. fellatio_rec [M] Receiving oral sex (fellatio; getting a blowjob)
  9. fellatio_to Giving a man oral sex (fellatio; giving a blowjob)
  10. cunn_rec [F] Receiving oral sex (cunnilingus; being eaten out)
  11. cunn_to Giving a woman oral sex (cunnilingus; eating her out)
  12. anil_to_f Giving a woman a rimjob (anilingus; licking a woman's anus)
  13. anil_to_m Giving a man a rimjob (anilingus; licking a man's anus)
  14. anil_rec Receiving a rimjob (anilingus; having your anus licked)
  15. mast_rec_m [M] Receiving masturbation from a partner (getting a handjob; being jacked off)
  16. mast_to_m Masturbating a man (giving a handjob; jacking him off)
  17. mast_rec_f [F] Receiving masturbation from a partner (having your clitoris or vagina touched by your partner's hands)
  18. mast_to_f Masturbating a woman (touching a woman's clitoris or vagina with your hands)
  19. objpen_rec Being penetrated by an object, such as a dildo
  20. objpen_to Penetrating your partner with an object, such as a dildo
  21. semen_to_mouth [M] Ejaculating (cumming) in your sex partner's mouth without a condom
  22. semen_to_vag [M] Ejaculating (cumming) in your sex partner's vagina without a condom
  23. semen_to_anus [M] Ejaculating (cumming) in your sex partner's anus without a condom
  24. semen_rec_mouth Receiving a man's semen/cum in your mouth without a condom
  25. semen_rec_vag [F] Receiving a man's semen/cum in your vagina without a condom
  26. semen_rec_anus Receiving a man's semen/cum in your anus without a condom
  27. semen_swallow Swallowing a man's semen/cum
  28. control_to Totally controlling your sex partner
  29. control_rec Being totally controlled by your sex partner
  30. pain_to Causing pain to your sex partner
  31. pain_rec Having your sex partner cause you pain
  32. bondage_to Tying up your sex partner as part of sex (bondage)
  33. bondage_rec Being tied up by your sex partner as part of sex (bondage)
  34. urine_to Urinating on your sex partner as part of sex (watersports)
  35. urine_rec Being urinated on by your sex partner as part of sex (watersports)
  36. leather Using leather as part of sex
  37. insult_to Calling your sex partner insulting names during sex (e.g., "slut")
  38. insult_rec Being called insulting names during sex (e.g., "slut")
  39. feet Sex inolving your or your partner's feet
  40. group_sex Having sex with two or more people at once
  41. voyeur_known Watching other people naked, masturbating, or having sex, when they know that you're watching
  42. voyeur_secret Watching other people naked, masturbating, or having sex, when they don't know that you're watching
  43. uncon Sex with someone while they're unconscious (e.g., asleep or passed out from alcohol or a drug)
  44. rape_to Raping someone
  45. rape_rec Being raped
  46. porn_written Reading pornographic or erotic stories
  47. porn_visual Viewing pornographic or erotic videos or pictures
  48. porn_video_mf Watching a video of a man and woman having sex
  49. porn_video_ff Watching a video of two women having sex
  50. porn_video_mm Watching a video of two men having sex
  51. incest Sex with someone you're related to (a family member)
  52. in_rship Sex with someone you're in a relationship with
  53. friend Sex with a friend you aren't in a relationship with
  54. recent_acq Sex with someone you met a few hours ago
  55. stranger Anonymous sex with a total stranger
  56. hug Hugging
  57. kiss Kissing
  58. intimacy Feeling emotionally close to your sex partner
  59. love Feeling in love with your partner
  60. desired Being sexually desired by your partner
  61. pleasure_to Giving your partner sexual pleasure
  62. penis_big Sex with a man with a particularly large penis
  63. penis_small Sex with a man with a particularly small penis
  64. breasts_big Sex with a woman with particularly large breasts
  65. breasts_small Sex with a woman with particularly small breasts
  66. butt_big Sex with someone with particularly large buttocks
  67. butt_small Sex with someone with particularly small buttocks
  68. feet_big Sex with someone with particularly large feet
  69. feet_small Sex with someone with particularly small feet
  70. man Sex with a man
  71. man_fem Sex with a very feminine man
  72. man_mas Sex with a very masculine man
  73. man_trans Sex with a trans-man (i.e., someone assigned female sex by a doctor at birth, but who now identifies as male)
  74. woman Sex with a woman
  75. woman_fem Sex with a very feminine woman
  76. woman_mas Sex with a very masculine woman
  77. woman_trans Sex with a trans-woman (i.e., someone assigned male sex by a doctor at birth, but who now identifies as female)
  78. age_08 Sex with an 8-year-old
  79. age_14 Sex with a 14-year-old
  80. age_70 Sex with a 70-year-old
  81. rich Sex with a wealthy person with a prestigious profession
  82. rebel Sex with a rebel or rule-breaker (bad boy, bad girl)
  83. experienced Sex with a person with lots of previous sexual experience
  84. virgin Sex with a virgin
  85. persity_eh Sex with an extraverted, enthusiastic person
  86. persity_al Sex with a critical, quarrelsome person
  87. persity_ch Sex with a dependable, self-disciplined person
  88. persity_nh Sex with an anxious, easily upset person
  89. persity_oh Sex with a person who is complex and open to new experiences
  90. persity_el Sex with a reserved, quiet person
  91. persity_ah Sex with a sympathetic, warm person
  92. persity_cl Sex with a disorganized, careless person
  93. persity_nl Sex with a calm, emotionally stable person
  94. persity_ol Sex with a conventional, uncreative person};

References

Ariely, D., Loewenstein, G., & Prelec, D. (2006). Tom Sawyer and the construction of value. Journal of Economic Behavior and Organization, 60, 1–10. doi:10.1016/j.jebo.2004.10.003

Bogaert, A. F. (2006). Toward a conceptual understanding of asexuality. Review of General Psychology, 10(3), 241–250. doi:10.1037/1089-2680.10.3.241

Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon's Mechanical Turk: A new source of inexpensive, yet high-quality, data? Perspectives on Psychological Science, 6, 3–5. doi:10.1177/1745691610393980

Chasin, C. J. D. (2011). Theoretical issues in the study of asexuality. Archives of Sexual Behavior, 40(4), 713–723. doi:10.1007/s10508-011-9757-x

Chivers, M. L., Seto, M. C., Lalumière, M. L., Laan, E., & Grimbos, T. (2010). Agreement of self-reported and genital measures of sexual arousal in men and women: A meta-analysis. Archives of Sexual Behavior, 39(1), 5–56. doi:10.1007/s10508-009-9556-9

Ellis, L., Burke, D., & Ames, M. A. (1987). Sexual orientation as a continuous variable: A comparison between the sexes. Archives of Sexual Behavior, 16(6), 523–529. doi:10.1007/BF01541716

Eriksson, K., & Simpson, B. (2010). Emotional reactions to losing explain gender differences in entering a risky lottery. Judgment and Decision Making, 5, 159–163.

Forbes, M. K., Kotov, R., Ruggero, C. J., Watson, D., Zimmerman, M., & Krueger, R. F. (2017). Delineating the joint hierarchical structure of clinical and personality disorders in an outpatient psychiatric sample. Comprehensive Psychiatry. Advance online publication. doi:10.1016/j.comppsych.2017.04.006

Gangestad, S. W., Bailey, J. M., & Martin, N. G. (2000). Taxometric analyses of sexual orientation and gender identity. Journal of Personality and Social Psychology, 78(6), 1109–1121. doi:10.1037/0022-3514.78.6.1109

Gildersleeve, K., Haselton, M. G., & Fales, M. R. (2014). Do women's mate preferences change across the ovulatory cycle? A meta-analytic review. Psychological Bulletin, 140(5), 1205–1259. doi:10.1037/a0035438

Goldberg, L. R. (1990). An alternative "description of personality": The Big-Five factor structure. Journal of Personality and Social Psychology, 59(6), 1216–1229. doi:10.1037/0022-3514.59.6.1216

Goldberg, L. R. (2006). Doing it all bass-ackwards: The development of hierarchical factor structures from the top down. Journal of Research in Personality, 40(4), 347–358. doi:10.1016/j.jrp.2006.01.001

Gosling, S. D., Rentfrow, P. J., & Swann, W. B., Jr. (2003). A very brief measure of the Big-Five personality domains. Journal of Research in Personality, 37(6), 504–528. doi:10.1016/S0092-6566(03)00046-1

Gottfredson, L. S. (1997). Why g matters: The complexity of everyday life. Intelligence, 24(1), 79–132. doi:10.1016/S0160-2896(97)90014-3

Haslam, N. (1997). Evidence that male sexual orientation is a matter of degree. Journal of Personality and Social Psychology, 73(4), 862–870. doi:10.1037/0022-3514.73.4.862

Ipeirotis, P. G. (2010). Demographics of Mechanical Turk (Technical Report No. CeDER-10-01). New York, NY: New York University. Retrieved from http://hdl.handle.net/2451/29585

Kendall, M. G. (1945). The treatment of ties in ranking problems. Biometrika, 33(3), 239–251. doi:10.2307/2332303

Kotov, R., Krueger, R. F., Watson, D., Achenbach, T. M., Althoff, R. R., Bagby, R. M., … Zimmerman, M. (2017). The Hierarchical Taxonomy of Psychopathology (HiTOP): A dimensional alternative to traditional nosologies. Journal of Abnormal Psychology, 126(4), 454–477. doi:10.1037/abn0000258

Kuper, L. E., Nussbaum, R., & Mustanski, B. (2012). Exploring the diversity of gender and sexual orientation identities in an online sample of transgender individuals. Journal of Sex Research, 49(2, 3), 244–254. doi:10.1080/00224499.2011.596954

Meana, M. (2010). Elucidating women's (hetero)sexual desire: Definitional challenges and content expansion. Journal of Sex Research, 47(2, 3), 104–122. doi:10.1080/00224490903402546

Mock, S. E., & Eibach, R. P. (2012). Stability and change in sexual orientation identity over a 10-year period in adulthood. Archives of Sexual Behavior, 41(3), 641–648. doi:10.1007/s10508-011-9761-1

Ogas, O., & Gaddam, S. (2011). A billion wicked thoughts: What the world's largest experiment reveals about human desire. New York, NY: Dutton. ISBN 978-0-525-95209-1.

Ott, M. Q., Corliss, H. L., Wypij, D., Rosario, M., & Austin, S. B. (2011). Stability and change in self-reported sexual orientation identity in young people: Application of mobility metrics. Archives of Sexual Behavior, 40(3), 519–532. doi:10.1007/s10508-010-9691-3

Rosenthal, A. M., Sylva, D., Safron, A., & Bailey, J. M. (2012). The male bisexuality debate revisited: Some bisexual men have bisexual arousal patterns. Archives of Sexual Behavior, 41(1), 135–147. doi:10.1007/s10508-011-9881-7

Savin-Williams, R. C., Joyner, K., & Rieger, G. (2012). Prevalence and stability of self-reported sexual orientation identity during young adulthood. Archives of Sexual Behavior, 41(1), 103–110. doi:10.1007/s10508-012-9913-y

Savin-Williams, R. C., & Vrangalova, Z. (2013). Mostly heterosexual as a distinct sexual orientation group: A systematic review of the empirical evidence. Developmental Review, 33(1), 58–88. doi:10.1016/j.dr.2013.01.001

Scorolli, C., Ghirlanda, S., Enquist, M., Zattoni, S., & Jannini, E. A. (2007). Relative prevalence of different fetishes. International Journal of Impotence Research, 19(4), 432–437. doi:10.1038/sj.ijir.3901547

Seto, M. C. (2012). Is pedophilia a sexual orientation? Archives of Sexual Behavior, 41(1), 231–236. doi:10.1007/s10508-011-9882-6

Suri, S., & Watts, D. J. (2011). Cooperation and contagion in web-based, networked public goods experiments. PLOS ONE. doi:10.1371/journal.pone.0016836

Toates, F. (2009). An integrative theoretical framework for understanding sexual motivation, arousal, and behavior. Journal of Sex Research, 46(2, 3), 168–193. doi:10.1080/00224490902747768