Concordance and Discordance between Paraphilic Interests and Behaviors: A Follow-Up Study

ABSTRACT

Although paraphilic interests represent significant risk factors for recidivism among sexual offenders, little is known about the magnitude of concordance between paraphilic interests and behaviors in the general population. The goal of this follow-up study was to conduct secondary analyses based on a sample of 1040 adults (475 men; 565 women) recruited in the general population. Levels of associations and active concordance (having both interest and experience), passive concordance (having neither interest nor experience), active discordance (having experience without interest) and passive discordance (having interest without experience) between paraphilic interest and corresponding behavior were assessed. Concordance and discordance indexes were also computed, as well as regressions and moderation analyses. As expected, paraphilic interests predicted corresponding behaviors, although the mean active concordance rate was only approximately 50%. Concordance rates varied with gender and the criminal nature (legal vs. illegal) of paraphilia. Paraphilic interests in adults from the general population may not have the same predictive value as that observed in medico-legal contexts. The possible role of other moderators in the concordance between paraphilic interest and behavior in non-clinical populations should be assessed. These findings have implications for sexual abuse prevention programs aiming at individuals in the community.

Introduction

In forensic psychology/sexology, sexual interests (fantasies, desires, urges) or preference for illegal behaviors represent clear (and rather evident) risk factors for recidivism (Hanson & Morton-Bourgon, 2005; Mann et al., 2010). It comes as no surprise, for instance, that men who have sexually offended against children are more likely to report pedophilic sexual fantasies than other men (e.g., Grundmann et al., 2016; Turner-Moore & Waterman, 2017). Consequently, most theories of sexual offending include sexual interests involving non-consenting behaviors as significant risk factors for the commission of that behavior (e.g., Abel & Blanchard, 1974; Prentky et al., 1989; Ward & Siegert, 2002; see Bartels & Gannon, 2011 for a review). However, presuming that the same level of concordance between paraphilic interests and behaviors applies to the general population might have serious negative consequences. For instance, with the increasing number of community programs aiming at the prevention of sexual abuse among the general population worldwide (e.g., Dunkelfeld in Germany, Beier et al., 2009; Prevent-Tell in Sweden, Oberg et al., 2016; S.T.O.P. in France, CRIAVS (Centres des Ressources pour les Intervenants auprès des Auteurs de Violences Sexuelles, 2021), it becomes crucial to avoid false-positive reports of potential offenders based solely on the nature of sexual interests. In addition, the definition of “imminent danger” underlying mandatory reports of clients at risk of offending by health care professionals is unclear in most jurisprudences (especially in non-forensic contexts), rendering it difficult (and uncomfortable) for clinicians to decide to produce a report based uniquely on sexual interests (Stephens et al., 2021). Clearly, a better understanding of the concordance vs. discordance between paraphilic sexual interests and associated behaviors among the general population is warranted. As stressed by the leaders of The Dunkelfeld Project about persons with paraphilic interests, “there are those who keep their desires in their fantasies and do not act them out on the behavioral level, [the number of which] can be increased by preventive therapeutic measures” (Beier, 2021, back cover).

Currently, little is known about the magnitude of concordance between paraphilic interests and paraphilic behaviors in the general population. Available reports of a link between sexual interests and corresponding behaviors typically involve sexual offenders without non-offender comparison groups (e.g., Baić et al., 2019; Chan et al., 2015; Gee et al., 2004; Skovran et al., 2010; Woodworth et al., 2013). As for studies conducted in the general population, they report relatively high rates of interest for paraphilic behaviors, although concordance between these interests and the corresponding behaviors is rarely assessed (Ahlers et al., 2011; Castellini et al., 2018; Dawson et al., 2016; Engel et al., 2019; Joyal & Carpentier, 2017; Joyal et al., 2015). However, the prevalence of paraphilic (and bondage/discipline, dominance/submission, sadism/masochism; BDSM) interests in the general population is usually higher than that of corresponding behaviors (Ahlers et al., 2011; Holvoet et al., 2017; Joyal & Carpentier, 2017). The few studies that directly assessed concordance between paraphilic interests and behaviors in the general population found positive significant associations, as expected, although with fluctuating magnitudes (ranging from medium to relatively large correlations, usually between .30 and .65; Bártová et al., 2021; Dombert et al., 2016; Knack, 1982; Noorishad et al., 2019; Person et al., 1992; Seto et al., 2021; Williams et al., 2009). One issue with associations between paraphilic interests and behaviors is that most scales include “no interest” and “no behavior” as response choices (e.g., Bártová et al., 2021). Given that this two-choice pairing (no interest and no behavior) is selected by the majority of respondents, passive concordance inflates correspondence between interests (none) and behaviors (none). To avoid this problem, associations between paraphilic interests and corresponding behaviors restricted to respondents who report interest should also be assessed. In addition, it is worth noting that high odds ratios for indulging in paraphilic behaviors based on the presence of a corresponding fantasy reflect only moderate concordance in the general population, i.e., approximately 50% of respondents report both fantasies and experience with the behavior (Dombert et al., 2016; Långström & Seto, 2006; Noorishad et al., 2019). In brief, these studies suggest that paraphilic interests are significantly associated with, but not necessarily predictive of, corresponding behaviors in the general population. It seems that non-negligible proportions of persons who have experience with paraphilic behaviors do not have corresponding sexual interests, whereas some proportions of persons with paraphilic interests have not acted on them. These proportions remain to be determined.

Based on a nearly representative sample of participants, we previously confirmed that the prevalence of paraphilic interests is higher than that of paraphilic behaviors in the general population (Joyal & Carpentier, 2017). However, as underlined by Seto et al. (2021), we did not assess the magnitude of associations and rates of concordance (and discordance) between paraphilic interests and behaviors in that report. Given the paucity and importance of this type of data, we conducted secondary analyses to address that limitation. The main objective of this follow-up report was to assess the relative levels of associations and rates of concordance (and discordance) between paraphilic interests and behaviors based on the same sample of participants. The secondary goal of this study was to evaluate the magnitude of four different types of correspondence between paraphilic interests and paraphilic behaviors: 1) active concordance (having both paraphilic interests and paraphilic experience); 2) passive concordance (reporting neither paraphilic interests nor paraphilic experience); 3) active discordance (having experienced paraphilic behaviors without interest), and; 4) passive discordance (having paraphilic interests without acting on them). These data should be useful not only to better understand the link between paraphilic interests and behaviors among adults from the community, but also for clinicians involved in prevention programs against sexual abuse targeting the general population. The third goal of this study was to compare concordance rates of legal vs. criminal paraphilias. It was first hypothesized that concordance rates for illegal paraphilias would be significantly lower than those of legal paraphilias in the general population because they represent criminal acts (Seto et al., 2021). Given that sexual interests are less closely associated with corresponding behaviors in women than in men (at least for sexual fantasies; Joyal et al., 2015), we also hypothesized that gender would play a significant predictive role in rates of endorsement of paraphilic behaviors and magnitude of correspondence between interests and behaviors (Seto et al., 2021).

Method

The sample consisted of 1,040 adults (475 men; 565 women) aged between 18 and 64 years who matched as closely as possible the corresponding population of the province of Québec (see Joyal & Carpentier, 2017 for details). Participants were asked about their interest in (“Would you like to … ”) and experience with (“Have you ever been sexually aroused by doing … ”) the main paraphilic behaviors of the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5, American Psychiatric Association, 2013; fetishism, transvestism, voyeurism, exhibitionism, frotteurism, sexual masochism, sexual sadism, and pedophilia). For each item, the degree of interest (0 = Not at all; 1 = I have thought about it; 2 = Maybe; 3 = Absolutely) and the frequency of behavior (0 = I have never done it; 1 = I did it once; 2 = I did it sometimes, 2–10 lifetime; 3 = I did it often, more than 10 times lifetime) were assessed.

To evaluate the magnitude of associations, a series of two-way Kendall’s Tau-B correlations (e.g., Bártová et al., 2021) were conducted between the level of paraphilic interest (ratings between 0 and 3) and the frequency of corresponding behavior (ratings between 0 and 3) for the whole sample. To assess the magnitude of concordance and discordance, paraphilic interests were dichotomized between Not at all (rating of 0) and At least some interest (ratings of 1 to 3) and paraphilic behavior was dichotomized between Not at all (rating of 0) and At least once in lifetime (ratings of 1 to 3). Based on these variables, four different types of correspondence were computed: active concordance (ratings of 1 to 3 on both interest and experience), passive concordance (ratings of 0 on both interest and experience), active discordance (ratings of 1–3 on experience and 0 on interest), and passive discordance (ratings of 1–3 on interest and 0 on experience). These analyses generated percentages of participants belonging to each type of concordance/discordance. All rates were computed for the entire sample and separately for men and women. An index of active concordance was also calculated, defined as the proportion of participants with both interest and experience for a given paraphilic behavior divided by the proportion of those who reported interest (with or without experience) for the same behavior. This index is an indicator of the magnitude of concordance, that is, the correspondence between having a paraphilic interest and having acted on it. For instance, an index of .50 indicates that only half of persons with interest in a given behavior also had experience with that behavior. Similarly, an index of active discordance was calculated, defined as the proportion of participants without interest but with experience for a given paraphilic behavior divided by the proportion of participants with both interest and experience for the same behavior. This index indicates the rates of participants who have done a paraphilic behavior without having interest for that behavior.

Finally, the potential effects of paraphilic criminal status (legal vs. illegal) and gender of participants (men vs. women) were assessed in line with Seto et al. (2021). First, associations between the level of interest and the level of the corresponding behavior (continuous variables) were compared between legal vs. illegal paraphilias with the Steiger’s (1980) test of mean correlation comparisons using the cocor package for R (Diedenhofen & Musch, 2015). Second, regression equations were conducted for each paraphilia with behavior as the dependent factor and both interest and being a man as predictors (Seto et al., 2021). Gender was then used as a moderator in associations between the presence of interest and the corresponding behavior for each paraphilia based on moderation analyses (Hayes, 2018) using the PROCESS macro v.3.5 for SPSS 28 (https://www.processmacro.org/index.html).

Results

As expected, correlations between paraphilic interests and corresponding behaviors were all positive and significant, ranging from .40 to .71 (Table 1). When only participants who reported interest were considered, most of these rates decreased, suggesting that the no-interest/no-behavior pairing explained a non-negligible part in these associations. Still, two of these rates increased, indicating a stronger association between level of interest and frequency of behavior when only persons with interest are included (sexual sadism and exhibitionism; Table 1).

Table 1. Correlations and regressions between paraphilic interests and correspondent behaviors

	τ	Model R^2	Interest	Gender (♂)	Intercept (Interest→♂→Behavior)
Voyeurism	.400	R^2 = .232. F = 103.43	B = .339	B = .349	B = .023
With interest^a	.197			p < .001	p = .63
Fetishism	.471	R^2 = .312, F = 155.15	B = .550	B = .130*	B = .115*
With interest	.384			p = .032	p = .013
Frotteurism	.577	R^2 = .378, F = 206.85	B = .575	B = .066	B = .026
With interest	.309			p = .14	p = .59
Masochism	.709	R^2 = .577, F = 471.04	B = .628	B = – .057^b	B = .043
With interest	.473			p = .08	p = .23
Sadism	.553	R^2 = .447, F = 278.13	B = .536	B = .017	B = .021
With interest	.650			p = .35	p = .60
Transvestism	.495	R^2 = .339, F = 175. 63	B = .661	B = .015	B = .370
With interest	.323			p = .43	p < .001
Exhibitionism	.444	R^2 = .234, F = 104.83	B = .496	B = .044*	B = .290
With interest	.516			p = .012	p < .001
Pedophilia	.611	N/A
With interest	.364

Bold: significant at p < .001. *Significant at p < .05. N/A: Not applicable due to the low rate of behaviors.

^aCorrelations between interest and behavior limited to participants who reported interest.

^bA negative sign (-) indicates that being a women was a better predictor.

As shown on Table 2, indexes of active concordance were also generally in line with those of previous studies (Dombert et al., 2016; Långström & Seto, 2006), i.e., averaging approximately .50. Again, these indexes varied across paraphilia subtypes, ranging from .42 (transvestism) to .67 (masochism and frotteurism). They also varied across genders, tending to be higher in men compared with women, with the notable exception of masochism (with an index of .71 for women). The two highest active concordance rates were found for voyeurism and frotteurism in men and masochism and frotteurism in women (Table 2). A minority of participants were also actively discordant (having experience without interest). As expected, indexes of active discordance (rates of engagement in a behavior without having interest for it) were generally lower than those of active concordance (rates of engagement in a behavior with interest for it). One exception was exhibitionism, for which indexes of discordance were higher than those of concordance (and higher than 1) for both genders. The lowest indexes of active discordance (.15 or less) were found for masochism (both genders) and fetishism (men).

Table 2. Rates of concordance and discordance between interest and experience for paraphilic behaviors

	Voyeurism.	Fetishism	Frotteurism	Masochism	Sadism	Transvestism	Exhibitionism	Pedophilia
Interest/experience	%	%	%	%	%	%	%	%
Have interest in	46.3	44.5	26.7	23.8	7.1	6.3	4.5	0.6
Interest without experience (passive discordance)	19.1	23.2	8.6	7.8	3.5	3.2	2.3	0.2
Experience without interest (active discordance)	8.8	5.1	7.8	2.3	2	1.8	2.8	0.1
Interest and experience (active concordance)	25.5	21.5	17.9	16	3	2.3	1.8	0.3
No interest and no experience (passive concordance)	46.6	50.2	65.7	73.9	91.5	92.7	93.1	99.4
Active concordance index ^a	.55	.48	.67	.67	.46	.42	.43	.50
Active discordance index ^b	.35	.24	.44	.14	.42	.37	1.6	.50
Men (n = 475)
Have interest in	60.0	40.4	34.2	19.2	9.5	7.2	5.9	1.1
Interest without experience (passive discordance)	19.1	22.6	11	7.4	4.9	3.8	2.3	0.4
Experience without interest (active discordance)	11.4	1.6	9.3	1.7	3	1.3	4.2	0.2
Interest and experience (active concordance)	40.2	25.8	23.3	11.6	4	3.6	3.2	0.4
No interest and no experience (passive concordance)	29.3	50	56.4	79.3	88.1	91.3	90.3	99.0
Active concordance index	.68	.53	.68	.61	.45	.49	.58	.36
Active discordance index	.28	.06	.40	.15	.75	.36	1.6	.50
Women (n = 565)
Have interest in	34.7	47.9	20.7	27.8	5.1	5.5	3.4	0.2
Interest without experience (passive discordance)	19	23.6	6.5	8.1	2.3	2.8	2.3	0
Experience without interest (active discordance)	6.7	3.5	6.5	2.8	1.1	2.3	1.6	0.2
Interest and experience (active concordance)	13.2	17.9	13.5	19.6	2.3	1.0	0.7	0
No interest and no experience (passive concordance)	61.1	55	73.5	69.5	94.3	93.9	95.4	99.6
Active concordance index	.41	.43	.68	.71	.50	.26	.23	N/A
Active discordance index	.51	.20	.48	.14	.49	2.3	2.3	N/A

N = 1040. N/A: Not applicable

^aActive concordance index: Have interest and experience/Have interest proportion.

^bActive discordance index: Have experience without interest/Have experience and interest proportion.

In accordance with the first hypothesis, the mean correlation between interests and behaviors was significantly higher (z = 2.0362, p < .05) for legal (.46) than for illegal paraphilia (.35). As expected, having the desire to engage in a paraphilic behavior represents a significant factor to realize it for all paraphilias (Table 1, see also Seto et al., 2021). Also in line with the report of Seto et al. (2021), the two highest associations between behaviors and interests were sexual masochism and sexual sadism (Table 1). In partial accordance with the second hypothesis, some (but not all) paraphilic behaviors were significantly associated with gender: fetishism (men), voyeurism (men), exhibitionism (men) and masochism (a trend toward women; Table 1). As for concordance between interest and behavior, it was significantly moderated by gender (being a man) for three paraphilias: fetishism, transvestism, exhibitionism (the number of pedophilic acts reported was too low to allow further statistical analyses).

Discussion

The main goal of this study was to evaluate the magnitude of association, concordance and discordance between interests for, and experience with, paraphilic behaviors in a sample of participants recruited among the general population. In line with previous studies (Bártová et al., 2021; Seto et al., 2021), correlations between the level of paraphilic interests (including none) and the frequency of corresponding behaviors (including none) were all positive and highly significant. When only participants with interest were considered, most of these associations decreased, indicating that the no-interest/no-behavior pairing explained in part these links. Still, all associations remained significant, although variation between them varied considerably (from a low of .197 for voyeurism to a high of .650 for sexual sadism). These results suggest that the link between interest and behavior fluctuates significantly across types of paraphilia. For instance, the three strongest associations between interest and behavior (all above .400) were found for sexual sadism, exhibitionism, and sexual masochism. It is worth noting that Seto et al. (2021) reported higher correlations between interest and behavior for masochism (.664) and sadism (.665) compared with other paraphilias. In addition, two rates in the present study increased when only interested participants were considered (sexual sadism and exhibitionism), indicating that for some paraphilias, associations between levels of interest and frequencies of behavior are in fact higher when absence of interest is eliminated from the equation. In these cases, the presence of interest may have a better predictive value for the appearance of the corresponding behavior.

Regression analyses also confirmed that having the desire to engage in a paraphilic behavior was significantly associated with acting on that behavior (see also Seto et al., 2021). Still, gender also represented a significant factor for the realization of some (but not all) paraphilic behaviors. As expected, fetishistic, voyeuristic, and exhibitionistic behaviors were associated with being a man and masochistic behaviors were (almost significantly) associated with being a woman. Seto et al. (2021) also found that four paraphilic behaviors were associated with gender: eroticized gender (men), frotteurism (men), zoophilia (men), and masochism (women). Given that the present study was not based on the same instrument and the same paraphilias, results are difficult to compare with those of Seto et al. (2021). Still, it is worth noting that in both studies, sexual masochism was associated with being a woman.

As for the more specific measure of correspondence between paraphilic interests and behaviors, the active concordance index, it was moderate on average (approximately .50), suggesting that about one person out of two with such interest also reported (or not) the corresponding behavior. Therefore, the presence of a paraphilic interest (or even a desire, as in this study) among people of the general population is only partially indicative of their experience with the corresponding behavior, perhaps less than what is usually reported in forensic contexts (e.g., Hanson & Morton-Bourgon, 2005; Mann et al., 2010). Concordance rates also vary significantly across paraphilias. A first source of variation was the criminal status (legal vs. illegal) of the paraphilia. Although the concordance rates suggest that the sole criminal nature of a paraphilia is not sufficient to explain the odds of acting out (e.g., voyeurism, frotteurism), legal paraphilic interests were significantly more likely to be associated with corresponding behaviors than interests for illegal sexual behaviors. A similar result was reported by Seto et al. (2021).

A second source of concordance variation was gender (moderation variable), although it was significant for only three paraphilias (fetishism, transvestism, and exhibitionism, all related with being a man). These results suggest that even if odds of engaging in paraphilic behaviors are generally higher in men, the link between having a paraphilic interest and realizing it is not necessarily weaker in women. Seto et al. (2021), for instance, found that pedohebephilic interests were significantly associated with pedohebephilic behaviors in both men and women (if heterosexual) among the general population. In the present study, women who had interests in sexual masochism, sexual sadism, frotteurism, or voyeurism were as likely as men to have experienced it.

The active discordance index also generated interesting results. For instance, correlations between interest and behavior were higher for exhibitionism after participants with no interest and no experience with exhibitionism were removed from the equation. These results suggest that having interest in exhibitionism is significantly associated with behavioral exhibitionism. However, the active discordant index indicated that exhibitionism behaviors were more likely to be reported by participants who engaged in these behaviors without interest in doing so (as indicated by indexes higher than 1). Therefore, among the general population, having interest for a paraphilia such as exhibitionism seems to be significantly associated with the corresponding behavior, whereas the opposite is not true, i.e., exhibitionism is not necessarily indicative of a corresponding interest. In opposition, indexes of active discordance were especially low (.15 or less) for masochism (men and women) and fetishism (men). These results suggest that most people from the general population who engage in these behaviors also have the desire to do so.

Future investigations should assess the presence and importance of other possible moderating factors underlying the variation of concordance between paraphilic (or any) sexual interest and behavior. For instance, Bondü and Birke (2021) recently reported a significant association between coercive sexual fantasies and behavioral sexual sadism (consenting and non-consenting) in young adults (mainly university students), although less than 25% of the variance was explained by the sole presence of these fantasies. Importantly, inclusion of other factors such as personality traits, attitudes, and general aggressiveness increased that percentage to nearly 50% (Bondü & Birke, 2021). At the individual level, variables such as sexual drive, sexual orientation, behavioral impulsivity, psychopathic traits, substance abuse, intelligence and educational levels deserve further attention (e.g., Bondü & Birke, 2021; Seto et al., 2021; Williams et al., 2009). At the sexual interest level, low diversity of interests (e.g., having few but highly specific fantasies) paired with high intensity (e.g., very sexually arousing and recurrent), a non-volitional type of appearance (obsessional), a long history of emergence (e.g., early adolescence) and/or high frequency represent good possible moderating factors in the odds of acting on them. Because the present study was not designed to assess concordance between interest and behaviors, these intermediate variables were not assessed.

Finally, different definitions and measures of sexual interest might explain in part variation in rates of concordance across studies. Fantasies about (Bártová et al., 2021), arousal associated with (Bártová et al., 2021; Seto et al., 2021), and desire to accomplish (current findings) a given sexual behavior can all be defined as sexual interest, although their proximity with behaviors might differ. Given that sexual fantasies are not necessarily associated with an actual desire to realize them (Joyal & Carpentier, 2017), concordance between behaviors and desire might be somewhat higher than that with fantasies. The definition of behavior is also important, such as including (or not) exclusive pornography use, which influences reported rates (see Dombert et al., 2016 for instance). Rates of pedophilic interests and experience (involving children aged 13 or less) were particularly low in this study (0.6% and 0,4%, respectively, compared with 4.1% and 3.2% among German men; Dombert et al., 2016), perhaps due to fear of being reported in Canada and/or the exclusion of behaviors limited to pornography consumption. In contrast, rates of interest and experience (Joyal & Carpentier, 2017) and concordance (this study) for frotteurism were relatively high in this sample. This reflects, at least in part, an important difference between our definition of frotteurism (“Touching or rubbing yourself against a stranger”) and that of other studies (“Touching or rubbing against a nonconsenting person,” e.g., Seto et al., 2021). Therefore, measures of paraphilic interests and behaviors should be validated for future studies, not only for convergent validity and reliability (Seto et al., 2021), but also for face and content validity.

Overall, there are growing data available concerning concordance and discordance between paraphilic interests and paraphilic behaviors among non-clinical, non-forensic samples of participants. Although there remain gaps in the literature, these investigations will help in developing risk assessment instruments for the general population, crucial for the emerging development of prevention programs against sexual abuse aiming at the community.

Disclosure Statement

No potential conflict of interest was reported by the author(s).