![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
ABSTRACT
Dominant/submissive role-play (D/s) is associated with specialized roles including Mistress, Master, Slave, Switch, Sadist, and Masochist. The current study uses cluster analysis to provide empirical evidence that no binary opposition or single spectrum constitutes a workable typology of individuals based on their affinities for these roles. The optimality of a particular choice of clustering scheme, including the number of clusters, is established using a replication technique which is presented in detail. A large number (n = 236,353) of individualized results (profiles) generated by the BDSM Test, a popular anonymous web survey, were analyzed. We hypothesize a two-dimensional typology of D/s profiles as the inferential result of our cluster analyses.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. We avoid writing this phrase the way it is usually stylized (“Dominant/submissive”) purely for the sake of typographical consistency. We capitalize the names of all kink roles. We use lowercase “dominant” and “submissive” only as descriptive adjectives. Gendered pairs of roles that denote the same functions when gender is set aside (e.g. Master and Mistress) are combined with a slash (Master/Mistress). See “Appendix” for our definition of “kink.”
2. See e.g. (Tukey, Citation1962).
3. The distinction between clusters and types of cluster has been adopted from (Lorr, Citation1983): “When a cluster analyst asks what a cluster means, what it represents, or what it implies, he or she is asking what construct or type underlies the data. A type is a construct inferred from the cluster.” Our definition—that clusters are of the same type if and only if they have similar statistical properties—differs from Lorr’s mainly in that it is explicitly quantitative.
4. In the case of hierarchical cluster analysis, where the output is a binary tree diagram illustrating how the dataset has been clustered into any number (between 1 and n) of disjoint clusters, some particular number J of clusters must nonetheless be chosen to obtain a particular partition into well-defined disjoint .
5. This may partly explain why many researchers do not fully explain their criteria for picking a clustering scheme, including the number of clusters, as (Clatworthy, Buick, Hankins, Weinman, & Horne, Citation2005) disapprovingly notes.
6. Here and throughout, we identify each survey result with a 25-dimensional vector. A set of survey results can then be regarded as a discrete set of points in 25-dimensional space. The boundary of a cluster typically refers to the topological boundary of the smallest convex set containing it.
7. hclust() in D. Müllner’s fastcluster package (version 1.1.25) for R.
8. gvlma() in Version 3.6.3 of R. See (Peña & Slate, Citation2006) for details about the GVLMA test, which incorporates the usual assumptions for linear regression.
9. The author has no affiliation with BdsmTest.org.
10. We use the gender-blind terms Bondage Top and Bondage Bottom where BdsmTest.org uses “Rigger” and “Rope Bunny,” respectively, in part because the definitions given on BdsmTest.org refer to not just to rope but to restraints in general, and in part to avoid associating the receptive role in bondage with a certain image of femininity.
11. Should we have wished to criticize the BDSM Test, we chose not to depend on its administrators for information. We consequently did not ask BdsmTest.org to share any data that was not publicly disclosed at the time of our writing, e.g. past questionnaires or scoring rubrics. All our information was obtained from publicly accessible profile pages.
12. Recall that Kendall’s τ depends only on ordinal comparisons of the form x1< x2, and that Kendall’s τb differs from Kendall’s τ in that it adjusts for ties.
13. Contrast with Paglia’s approach: “In researching sadomasochism, I did not begin with a priori assumptions […] I let the evidence suggest the theories. My conclusion, after wide reading in anthropology and psychology, was that sadomasochism is an archaic ritual form that descends from prehistoric nature cults and that erupts in sophisticated ‘late’ phases of culture” (Paglia, Citation2013). We strongly sympathize with the sentiment initially expressed in the quote. But the identification of modern-day “sadomasochism” with various practices spanning the history and prehistory of humanity belies the claim that no prior assumptions were made. In the current study we did not begin by assuming that sadism and masochism are manifestations of a single underlying psychological condition (Freud, Citation1975) or a single intersubjective relation (Lacan, Citation1991), or that sadism and masochism are two irreconcilable projects (Deleuze, Citation1989), or that sadism and masochism are linked to a unity of human experiences ranging from the present day to time immemorial. To accept or reject any of these premises prior to EDA would have violated the logic of our method. Rather, we took pains to let historically localized empirical data determine our hypothesized explanatory model. Crucially, in doing so, we viewed the data as archival evidence of social and discursive facts rather than anthropological or psychological ones (see next subsection, “Sociological value of the dataset”).
14. We have in mind a strain of social analysis which might be called poststructural and functionalist. It is straightforward to interpret the three “postulates in functional analysis” identified as “prevailing” in (Merton, Citation1968) (functional unity of society, universal functionalism, and “indispensability” of every element) in a way that is consistent both with Foucauldian archeology (Deleuze, Citation1988) and Deleuzoguattarian libidinal-social ontology (Deleuze & Guattari, Citation1996, Citation1998), provided the word “unity” is replaced by “multiplicity”—an uncontroversial modification in the post-“organismic” (Merton, Citation1968) era of sociology.
15. The traditions of “Old Guard” Leather exemplify this phenomenon. Newcomers are expected to master standardized roles, including by seeking mentorship (Lords, Citation2021).
16. The BDSM Test’s definitions of kink roles can be accessed at https://bdsmtest.org/info without taking the survey.
17. The precise distinctions between the kink roles Ageplayer, Boy/Girl, and Brat may be obscure to many academics, but these terms have clear practical meanings for those who identify themselves and each other as such. These terms are time-tested, too. Some version of the currently existing kink community and its jargon has continuously existed at least since the Second World War: “before and during the war, kinky folks seeking to identify each other would sometimes defensively ask, ‘Do you play the mandolin or the saxophone?’ to discover which of them was the masochist or the sadist by the first letter of these instruments. […] The creation of a butch subculture by the gay vets began to allow people to specialize their sexual interests in a way that had been impossible earlier.” (Baldwin, Citation2003, emphasis added)
18. Cf. Deleuze’s commentary on Foucault’s “archaeological” method, e.g.: “For Foucault, regularity has a precise meaning: it is the curve joining individual points (a rule)” (Deleuze, Citation1988)—and Foucault’s own definitions of “archaeology,” e.g.: “It is these rules of formation […] that I have tried to reveal, […] at a level I have called, somewhat arbitrarily perhaps, archaeological.” (Foreword to English Edition, Foucault, Citation1973)
19. Our working definition of discourse is that it is made up of statements which, when taken together as a totality, differentiate and assign roles and functions: “a statement […] is always part of a network of statements, in which it has a role, however minimal it may be, to play. […] There is no statement that does not presuppose others; there is no statement that is not surrounded by […] a distribution of functions and roles.” (Foucault, Citation1982)
20. What is “heuristic” in our technique is captured in the disjunction, “Either there exists an optimal clustering scheme with J clusters for some J such that 1 < J≪n, or not.” The Optimality Criterion is stated in terms of maximizing and minimizing over finitely many cases J ≤ n. Since the trivial cases J = 1 (no disaggregation) and J = n (every distinguishable individual an Island) are vacuously optimal, reasonable bounds must be imposed on the number of clusters. That is, it is required that 2 ≤ J≪n: the number of clusters must be greater than 1 and much less than the number of individuals. In practice, though, one does not typically begin by postulating an upper bound on J. Instead, the case J = 2 is tried, J is incremented until a satisfactory partition is found, and then a few more consecutive values of J are checked to see if adding a few clusters would improve the quality of the results. This is the first term of the disjunction (1 < J≪n). But one would in any case stop incrementing J well before reaching n, unless there were no workable cases with small J—that is, if all partitions with J≪n were unsatisfactory in the researcher’s judgment. This is the second term (“or not”).
21. When carrying out the Replication Procedure, we varied our choice of metric (L1, L2, L∞) as well as the number J∈{2, 3, 4, 5, 6} of clusters. The L1 metric with J = 4 produced the most closely packed and clearly distinct groupings of average vectors out of all choices of parameters tried. In all 15 cases we used Ward’s criteria (“method = ward.D2” for hclust() in R), having chosen it over single, average, and complete linkage methods by comparing the agglomerative coefficient AC obtained with all three linkage methods, all three metrics, and various subsample sizes ≤ 1000. Only Ward’s method consistently yielded AC>.9, an indication that “a clear clustering structure has been identified” (Kaufman & Rousseeuw, Citation1990).
22. Compare with inferential techniques of topological data analysis used e.g. in genomic analysis (Arsuaga, Citation2011) and DARPA-funded machine learning (Beyer, Citation2015).
23. E.g. Robert Coover’s novella Spanking the Maid (1982). Cf. also Simmel (here summarized by Coser) on domination: “domination does not consist in […] unilateral imposition of the superordinate’s will […] but, rather, […] involves reciprocal action.” (Coser, Citation1965)
24. The terms “D-type” and “s-type” (Masri, Citation2017) are often encountered in the community, where they are used (1) to classify roles as “Dominant-type” or “submissive-type” and (2) as terms for individuals who have affinities for such roles.
25. Recall that the output of a hierarchical clustering algorithm is a family of partitions indexed by the number J of clusters in each partition, and that the researcher must make a particular choice of J to obtain a particular partition.
26. E.g. by third-gendering trans people, who for a long time have sought refuge from harsh social treatment in the kink community, and whom for that reason alone it is plausible to assume may be overrepresented in kink-friendly spaces, especially online.
27. E.g. (Breslow, Evans, & Langley, Citation1985), (Breslow, Evans, & Langley, Citation1986), (De Neef, Coppens, Huys, & Morrens, Citation2019), (Herbenick et al., Citation2020), (Holvoet et al., Citation2017), (Joyal, Cossette, & Lapierre, Citation2015), (Joyal & Carpentier, Citation2017), (Jozifkova, Citation2018), (Jozifkova & Flegr, Citation2006), (Långström & Seto, Citation2006), (Levitt, Moser, & Jamison, Citation1994), (Martinez, Citation2018), (Moser & Levitt, Citation1987), (Nordling, Sandnabba, Santtila, & Alison, Citation2006), (Rehor, Citation2015), (Richters, De Visser, Rissel, Grulich, & Smith, Citation2008), (Rogak & Connor, Citation2018), and Sandnabba, Santtila, & Nordling, Citation1999).
28. One might ask what the “typical” gender and orientation of a Submissive is, for example. But we doubt that many researchers expect novel insights to emerge from predictive models which explain prevalence of sexual practices based on gender and orientation. It is not news that masculinity is more widely associated with dominance than submissiveness, etc. It is often worthwhile to question widespread prejudices where we find they are at odds with the data, to critique them on theoretical grounds, or to contest them as part of a political project. But simply to report that certain sexual practices are statistically “uncommon” among certain identity groups at best merely repeats common knowledge and at worst helps to marginalize exceptional individuals.
29. For example, increasing social acceptance of the transgender population has certainly diminished to some extent the tendency to view sexual attraction to trans people as a fetish rather than as a manifestation of “normal” sexuality.
30. We ask rhetorically whether sexual attraction to men with vaginas or women with penises is heteronormative, homonormative, or neither. No matter what we feel the answer should be to this sensitive question, the matter has certainly not been universally settled as of the time of our writing. Fetishism has often been associated with certain subsets of the transgender population—for example, by Freund and Blanchard, who characterized a certain “type” of trans women as committing an “error” in so far as they derive pleasure from believing themselves to be sexually attractive (Blanchard, Citation1989; Freund & Blanchard, Citation1993). This odious way of thinking survives in the minds of many now living. The point is that we cannot define “kink” simply to be coextensive with non-normative sexual practices. Different subpopulations maintain different norms. The collection of non-normative sexual practices is not well-defined.
31. The fact that non-normativity is a necessary condition for “kink” is illustrated by the cisgender comedian who said his kink is to climax during traditional intercourse with a woman. The joke is that traditional sexual intercourse is not “kinky”: it is not in itself non-normative. On the other hand, even the most socially acceptable sexual act can be made “kinky” if it is supplied with a suitable non-normative element, e.g. the ritualized suspension of a taboo, or an ironic or humorous “project” as described in Sade’s novels or Sacher-Masoch’s Mistress/Slave contracts (Deleuze, Citation1989; cf. Deleuze & Guattari, Citation1998, where masochistic “programs” are understood in terms of experimental “training”).
32. To be “kinky” is thus to be sexually outside the mainstream and inside the range of practices that are legible as “kink” within kink subcultures. This rules out coercive or criminal practices that violate the community’s ethical norms, e.g. the extraction of false consent under duress. (See Diane Vera's essay “Nine degrees of submission“ in (Califia, Citation1988), where real-world Mistress-Slave relationships are distinguished from “situations where the ‘consent’ is induced by brainwashing and/or social or economic pressures, and hence isn’t fully consensual”.) It also rules out sodomy whether or not it is locally unlawful. By contrast, the historical circumstances of gay male fisting clubs force us to regard fisting as “kink” despite the act itself being neither illegal nor uniformly non-normative for all combinations of gender and orientation. The distinction we have drawn here is between normative and non-normative practices (oral and anal sex vs. fisting) within gay male culture.
33. Our formal definition of “kink” stands or falls with the correlative definition of “the kink community.” If a cult coerces its members into viewing some criminal sexual behavior as “normative,” at least within the confines of the cult’s operations, should we understand the cult as being part of “the community”? If the larger society drives practitioners of some sexual practice into hiding, does that transform the practice into “kink”? It depends on what one means by “the” community. But we do not define the “kink community” formally. We do not mean just any community.
34. Here we of course have in mind Eco’s claims that fascism is “syncretistic” and can only be defined with a degree of “fuzziness”: its “features cannot be organized into a system; many of them contradict each other, and are also typical of other kinds of [related social forms—in Eco’s case,] despotism or fanaticism.” (Eco, Citation1995)
35. SM (or S/M, or S&M) stands for sadism and masochism. BDSM is generally understood as being short for Bondage, Discipline, Dominance, Submission, Sadism, and Masochism, with the middle letters doing double duty.
36. In “primal” play, participants literally wrestle for control, the administration of pain is neither a sufficient nor necessary condition, and restraints are not typically used. To call it “BDSM” therefore strikes us as technically imprecise, notwithstanding the fact that dominant and submissive roles (Hunter and Prey) can be identified. Applying the label of BDSM (let alone SM) to Feminizer/Sissy role-play is even more tendentious. According to Wikipedia Contributors (Citation2021), the present author among them, “[d]espite being labeled as ‘forced’ feminization […] not all participants are interested in BDSM aspects of the practice.”
37. Cf. (Cheng, Citation2020).
38. Cf. (Deleuze & Guattari, Citation1998): “It is, of course, indispensable for women to conduct a molar politics, with a view to winning back their own organism, their own history, their own subjectivity: ‘we as women … ’”