Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Neurodevelopmental Differences, Pedohebephilia, and Sexual Offending: Findings from Two Online Surveys

Neurodevelopmental Differences, Pedohebephilia, and Sexual Offending: Findings from Two Online... The neurodevelopmental theory of pedohebephilia states that sexual interests in children arise from early neurodevelopmental perturbations, as, for example, evidenced by increased non-right-handedness, more childhood head injuries, and reduced intelligence and height. As corroborating evidence largely rests on samples of convicted men, we conducted online surveys among German-speaking (Study 1, N = 199) and English-speaking men (Study 2, N = 632), specifically targeting community members with pedohebephilic or teleiophilic interests. Although we detected theoretically meaningful sexual interest pat- terns in an embedded viewing time task, we could not detect expected neurodevelopmental differences between teleiophilic and pedohebephilic men in either of the two studies. Strikingly, pedohebephilic men who reported convictions for sexual offenses emerged as shorter and less intelligent than pedohebephilic men without convictions in Study 2. While elucidating possible third variable confounds, results have to be interpreted cautiously because of the methodological problems inherent to non-matched case control designs. Keywords Pedophilia · Child sexual abuse · Neurobiology · Neurodevelopment · Etiology · DSM-5 Introduction Cantor et al., 2007, 2005a,b; Fazio, 2018). Yet, as the majority of studies are mostly restricted to samples of men who have In order to solve the puzzle of what may cause pedophilia or sexually offended, it is problematic that some of these factors hebephilia, that is, adult men's sexual attraction to sexually are also known markers of criminal behavior (e.g., lower height immature pre- or pubescent partners, respectively (Seto, 2012), and intelligence, Beckley et al., 2014; non-right-handedness, numerous studies have compared neurodevelopmental differ - Bogaert, 2001; early head injuries, Liu, 2011). Adding the fact ences between participants with and without pedophilia (see that there are many men with pedohebephilic interests who Tenbergen et al., 2015 for an overview). For example, neuroim- refrain from offending and are socially well adapted (Joyal aging techniques revealed structural or functional differences in et al., 2019), it is questionable whether these differences really the brains of pedophilic and teleiophilic (i.e., sexually attracted are due to patterns of sexual attraction and not to other fac- to adults) participants (e.g., Cantor et al., 2008; Poeppl et al., tors that are associated with (having been or) being arrested 2011, 2015). Furthermore, several studies found an association or convicted. To rule out such an alternative interpretation, a between pedophilia and soft markers for early neurodevelop- replication of neurodevelopmental differences between par - mental perturbations, such as non-right-handedness, height, IQ, ticipants with and without pedophilia in community samples or head injuries before age 13 (Blanchard et al., 2002, 2007; seems mandatory (Joyal et al., 2007). For solving the cardinal problem of recruiting non-insti- tutionalized participants with an uncommon and stigmatized * Sara Jahnke sexual attraction pattern such as pedophilia or hebephilia, the sara.jahnke@uib.no anonymity of online studies might be particularly helpful. Department of Health Promotion and Development, Following this logic, we sought to compare markers for neu- University of Bergen, 5020 Bergen, Norway rodevelopmental differences among men with different sexual Department of Psychology, Social & Legal Psychology, maturity interests in German-speaking (Study 1) and English- Johannes Gutenberg-Universität Mainz, Mainz, Germany speaking (Study 2) online samples. With a combined sample Institute for Clinical Psychology and Psychotherapy, of 831 participants, we hope to extend the currently limited Technische Universität Dresden, Dresden, Germany Vol.:(0123456789) 1 3 850 Archives of Sexual Behavior (2022) 51:849–866 evidence base concerning the relationship between neurodevel- to a pubescent and finally a postpubescent state (Blanchard opmental differences, pedohebephilia, and offending behavior. et al., 2007; see also Stephens et al., 2017 for a discussion of similar arousal patterns among hebephilic and pedophilic A Critical Discussion of the Early men who have sexually offended). Empirically, hebephilic Neurodevelopmental Perturbations Hypothesis men have sometimes been found to show IQ scores, rates of right-handedness, and childhood brain injuries that are Currently, pedohebephilic interest is increasingly viewed intermediate between those obtained among pedophilic and as being caused by an “underlying brain dysfunction, one teleiophilic men in forensic samples (Blanchard et al., 2003, that prevented the development of more typical intellectual 2007; Cantor et al., 2005a, 2005b). and sexual characteristics” (Cantor et al., 2005a, 2005b, p. 448). Neurodevelopmental deficits can be caused by genetic Potential Biases Associated with Studying disorders, accidents, brain tumors, and pre- or postnatal Pedohebephilia in Clinical or Forensic Samples exposure to teratogenic/toxic substances in utero (Becerra García, 2009). While a variety of such perturbations can Prior studies attempted to control for factors associated with have severe and long-lasting effects on cognitive function - criminality by comparing people with pedohebephilia with ing, they also manifest as "very mild, often inconsequential teleiophilic people, who have also mostly been referred features that are associated with atypical neurodevelopment" because of criminal or disturbing sexual activities (e.g., Blan- (Fazio, 2018, p. 1205), such as non-right-handedness, unde- chard et al., 2002; Fazio et al., 2017). Yet, just as increased tached ear lobes, or other biomarkers (see Jordan et al., 2020 levels of antisociality among teleiophilic men who have sexu- for a critical overview). Empirical research supporting the ally offended would not lead us to suppose that being anti- neurodevelopmental perturbations hypothesis of pedohebe- social is a characteristic feature of teleiophilia, the problem philia has mostly been conducted on men who have been remains that the detected markers for neurodevelopmental referred to clinical institutions "as a result of illegal or clini- perturbations may be more relevant for pedohebephilic men cally significant sexual behaviors or interests" (Cantor et al., who sexually offended than for pedohebephilia per se. Some 2007, p. 397). Based on those samples, findings from several researchers have therefore cautioned that "it seems far-fetched neuroimaging studies indicated structural abnormalities in to associate pedophilia with neurodevelopmental disorders" the brains of pedophilic men who have sexually offended (Joyal et al., 2019, p. 154) as long as unequivocal empirical compared to forensic or non-forensic teleiophilic controls support for such an account is lacking. (Cantor et al., 2008; Schiffer et al., 2007). Researchers have Additionally, some neurodevelopmental markers dis- also reported higher rates of non-right-handedness (Blan- cussed as being specific for pedophilia, such as increased chard et al., 2007; Cantor et al., 2005a, 2005b), lower height rates of left handedness/ambidexterity, have also been found (Cantor et al., 2007; Fazio et al., 2017), more head injuries among men with non-heterosexual orientation (Lalumière during childhood (Blanchard et al., 2002, 2003), more minor et al., 2000) or asexuality (Yule et al., 2014). Moreover, physical abnormalities (Dyshniku et al., 2015), and lower non-right-handedness has been linked to alcohol consump- IQs (Blanchard et al., 2007) among forensic/clinical samples tion (Denny, 2011) and a plethora of other psychological of pedohebephilic men. Notably, data on soft markers for disorders such as posttraumatic stress disorder (Boscarino neurodevelopmental perturbations especially need replica- & Hoffman, 2007 ), depression (Denny, 2009), or anorexia tion, because they were mostly obtained in the same insti- nervosa (Tenconi et al., 2010), to name a few. Lower IQ, tution (Centre for Addiction and Mental Health, Toronto) lower fraternal birth order, neuropsychiatric abnormalities, from members of the Kurt Freund Laboratory. All studies non-right-handedness, and lower D2/D4 length are also asso- were conducted on (potentially overlapping samples of) cases ciated with sexual offending in general, non-sexual delin- referred mostly for forensic evaluations (see, e.g., Blanchard quency, thrill seeking proclivity, and aggressive traits (Bailey et al., 2002), precluding the recruitment of larger numbers of & Hurd, 2005; Bogaert, 2001; Brower & Price, 2001; Fink pedohebephilic men who have not offended. et al., 2006; Moffitt et al., 1994; Ogunfowokan et al., 2016). Hebephilia is not a universally accepted concept, and the These findings underscore the unspecificity of the neurode- proposition to include sexual interests in pubescents as a dis- velopmental markers that have been associated with pedo- order category was not adopted in the DSM-5 (Singy, 2015). hebephilia, and invite speculation whether the differences Yet, irrespective of whether hebephilic interests can be con- between pedohebephilic men who have sexually offended and sidered pathological, the concept has demonstrated its merit teleiophilic control groups may be attributable to variables for research purposes. There are good reasons to assume that other than sexual interest. The same problems apply to stud- sexual attraction to prepubescents and to pubescents have ies that compare pedohebephilic men from treatment settings similar causes, since the course of sexual maturation leads to with healthy teleiophilic men, as the former group is likely gradual changes in physical appearance from a pre-pubescent to experience more distress about their sexuality or more 1 3 Archives of Sexual Behavior (2022) 51:849–866 851 concerns about sexual offending than pedohebephilic men The Present Studies who did not seek help. In a recent Swedish study, Abé et al. (2021) compared 55 help-seeking men with pedophilic dis- In the following studies, we sought to test selected indicators order from a national helpline with 57 age-matched controls of neurodevelopmental differences within two community recruited via the institute homepage and university services. samples, focusing solely on markers that are accessible in an The authors also report significant differences between the anonymous online setting. Prior studies on the neurodevelop- brains of pedophilic patients and controls, which, for the most mental correlates of pedophilia among forensic samples typi- part, remained significant when excluding patients who had cally classified participants as pedohebephilic either based on committed either sexual abuse or child pornography offenses, offense behavioral indicators as proxies (e.g., Cantor et al., indicating that these findings are not explained solely by 2005a, 2005b) or on participants’ responses in a phallometric criminality or antisociality. Furthermore, pedophilic disor- test or their self-reported sexual interests (e.g., Blanchard der was linked to a lower IQ, higher rates of several mood et al., 2002, 2003; Fazio et al., 2017). In the present studies, and anxiety disorders, and antisocial personality disorders, we employed self-report and viewing time (VT) measures to while the study found no differences regarding handedness assess sexual maturity interests. The VT measure was cho- and height (with pedophilic men emerging descriptively as sen as an objective measure (Schmidt et al., 2017) for cross- taller and less likely to be non-right-handed). validation of the self-report. By not relying on expert diagnosis of pedophilic disorder, Neurodevelopmental Perturbation Markers we avoided classifying participants based on categorical dis- in Community Samples of Pedohebephilic Men order diagnoses which in the case of pedophilia suffer from particularly high uncertainty (i.e., roughly between one- and About 50% or more of sexual offenses against children are two-thirds of pedophilia diagnoses may be considered wrong committed by teleiophilic men (Schmidt et al., 2013; Seto, due to low interrater reliability, Mokros et al., 2018). Our 2008), whereas many pedophilic men never commit sexual community data are also less likely to be confounded by third offenses at all (Dombert et al., 2016; Joyal & Carpentier, variables that are associated with caseness status. This is a 2021). Pedohebephilic men in community settings tend to typical problem of case–control designs, whereby individuals show better psychological adjustment than those recruited who by definition need to show clinically significant impair - in clinical or forensic environments (Jahnke et al., 2015). ment in terms of subjective distress or interpersonal function- Recently, an important first step toward disentangling fac- ing are compared to persons who are unlikely to exhibit signs tors associated with sexual offending and pedohebephilic of distress or functioning deficits. Yet, note that non-matched attraction has been made. Based on data from the German case–control studies such as the present ones are tainted by research project Neural Mechanisms underlying Pedophilia other potential biases, which we will discuss in limitations and Child Sexual Abuse (NeMUP), Gerwinn et al. (2018) section. found that "many of the factors reported as being related to paedophilia may […] actually be associated with commit- Hypotheses Above and beyond testing whether VT para- ting child sex offences and vice versa" (p. 75). In their sam- digms were valid in a specific online environment, we will ple, pedophilic and non-pedophilic men did not differ with test the following hypotheses based on the discussed limita- regard to handedness, accidents leading to unconsciousness tions of the empirical evidence for the neurodevelopmental in childhood, or IQ. Yet, with 155 pedophilic participants perturbation hypothesis: (about half of which had never committed sexual offenses), a larger evidence base is needed to achieve sufficient test (1) In case the neurodevelopmental perturbation hypoth- power to corroborate the purported small effect sizes of links esis will be specic fi for pedohebephilic sexual interests, between pedohebephilia and markers for neurodevelopmental pedohebephilic men will differ from teleiophilic men impairment, as the authors point out themselves. Regarding (i.e., men with a sexual interest in sexually mature indi- further indicators of neurodevelopmental differences, results viduals) with respect to markers of neurodevelopmental based on participants in the NeMUP project revealed simi- perturbations in community samples. lar results, namely that structural abnormalities in the brain (2) In case of a confound with self-reported sexual offend- (Schiffer et al., 2017) and impairments in executive function- ing status, pedohebephilic men who have sexually ing (Massau et al., 2017) are linked to offending status but offended will differ from pedohebephilic men who have not sexual interests (cf. Abé et al., 2021 who found that most not offended with respect to markers of neurodevelop- brain differences remained significant when controlling for mental perturbations. the sexual offending status of their pedophilic participants). We will test the following markers for neurodevelopmental perturbations: (a) height, (b) IQ, (c) non-right-handedness, 1 3 852 Archives of Sexual Behavior (2022) 51:849–866 Table 1 Sample size calculations Marker for neurodevelopmental deficits Source Effect size Comparison group Required sample size (two-tailed testing) Height McPhail and Cantor d = .20 Teleiophilic vs. pedohebephilic men 788 (2015) d = .21 Teleiophilic vs. pedophilic men (one-tailed: 620) (one-tailed: 564) Handedness Cantor et al. (2004) d = .25 Teleiophilic vs. pedohebephilic men 506 d = .50 Teleiophilic vs. pedophilic men 128 Head injuries before age 13 Blanchard et al. (2003) r = .12 None, pedophilic, hebephilic, and teleio- 542 philic participants were treated as an ordered set IQ (Wechsler Adult Intelligence Scale) Cantor et al. (2004) d = .32 Teleiophilic vs. pedohebephilic men 310 d = .53 Teleiophilic vs. pedophilic men 114 Calculations conducted with G*Power (for all calculations: 1–β = .80, α = .05, for height, handedness and IQ: t test, means: difference between two independent means, two groups: allocation ratio, N2/N1 = 1, for head injuries: exact, correlation, bivariate normal model: ρ H0 = 0, note that Spearman and Pearson correlations are computationally identical) Calculated from summary statistics in Table 1 and in-text (Cantor et al., 2004, p. 7), d = M –M / SD 1 2 pooled and (d) injuries resulting in unconsciousness before and after feasibility of the online design. Thus, we focused more on the age 13, all of which will be tested as separate hypotheses. descriptive point estimates of effect sizes than significance Study 1 was conducted among German-speaking individuals. levels. After we removed responses from 16 self-reported In Study 2, we improved the assessments of IQ and hand- female participants and two participants who did not report edness while repeating the study in a considerably larger their sex, the dataset contained 199 participants. Detailed English-speaking sample. information on how we categorized participants based on their sexual interests and offending status is presented in a separate section after the description of our measures. The Study 1 study was approved by the ethical review board of the Tech- nische Universität Dresden. Method Measures Participants and Procedure Participants filled out the scales in the order presented below. All participants were recruited on websites, blogs, and web We included additional questionnaires to assess interests forums to participate in a study on neurological develop- in sadomasochism, traumatic childhood experiences, and mental problems and traumatic childhood experiences. No non-negative sexual experiences in childhood that will be compensation was offered. Pedohebephilic individuals were featured in a separate publication (Jahnke et al., 2021). We sampled from websites for people with such sexual inter- were limited in our choice of intelligence measures, as most ests (jungsforum.net, schicksal-und-herausforderung.de, validated IQ scales are restricted to offline use by copyright krumme13.org, boylandonline.com, Deutsches Girllover- holders. Therefore, we used a scale to assess crystallized forum, ITP-Arcados), while teleiophilic participants were intelligence (Schipolowski et al., 2014) as the only validated recruited on science/psychology-related websites (e.g., German scale that was, to our knowledge, available for online forschung-erleben.uni-mannheim.de, caz-lesen.de, psychol- use at that time. We also developed a digit span test based ogie-heute.de, Facebook group "Psychologische Studien für on corresponding working memory tests in the Wechsler alle"). We attempted to collect as much information as pos- Adult Intelligence Scale (Lichtenberger & Kaufman, 2009). sible in a German-speaking sample. We stopped data col- Yet, because of severe validity concerns regarding these two lection when participation rates became very low. We did measures, we will not report their results. We calculated not attempt to reach a specific number of participants, as it The large majority (63%) of the participants scored between 10 and was clear to us prior to the study that a statistically sufficient 12 points on the crystallized intelligence test (note that the total score number of participants in German forums only will be unre- could theoretically range from 0 [no correct answer] to 12 [all items alistic (see Table 1 for sample size calculations). In contrast answered correctly]), indicating ceiling effects. Furthermore, a sizea- ble number of participants correctly recalled even 12-digit numbers on to the larger-scale Study 2, Study 1 was conducted to test the 1 3 Archives of Sexual Behavior (2022) 51:849–866 853 Cronbach’s α based on all of the 199 male participants in Head Injuries Participants indicated how many head injuries the dataset. leading to unconsciousness they had sustained before and after age 13. To prompt participant's memories, we trans- Viewing Time Measure of Sexual Interest We presented five lated and adapted questions from Blanchard et al. (2003), male and five female pictures of prepubescent children (Tan- replacing "hockey" with "soccer" as a more popular sports ner stage 1; Tanner, 1990) and postpubescent adults each activity among German children: "Were you ever knocked (Tanner stage 5) from the Not Real People Set (i.e., computer- unconscious during your childhood (before you were 13 years generated pictures of Caucasian individuals in bathing suits old), for example, falling from a tree or hitting your head in provided by the Pacific Psychological Assessment Coopera- a soccer game?" and "Were you ever knocked unconscious tion, 2004) in randomized order. Participants were prompted in adulthood (after 13 years of age), for example, in a car to indicate their attraction in a forced choice format with two accident or in a sports injury?". response alternatives within a one second response window: "Yes, this is a potential sexual partner for me" vs. "No, this is Social Desirability Tendencies to give socially desirable not a potential sexual partner for me" (i.e., speeded VT task, responses were assessed with an eight-item German scale by Imhoff et al., 2010, Study 3, but note that we did not exclude Ray (1984). Participants were asked to respond to questions trials with longer reaction times). We chose to instruct like “Are you quick to admit making a mistake?” with either speeded responding to prevent participants from producing "true" or "false." Higher values on the scale average indicate too many outliers due to unrestricted response windows in a higher risk for socially desirable responding (α = 0.67). an online survey. Response latencies were recorded unob- trusively. For further details on how we dealt with statistical Sociodemographic Information (Including Height and Previ‑ outliers see the section on VT classifications. Viewing time ous Convictions) We assessed participant sex (male, female, measures of pedohebephilia have repeatedly been shown to other), age, educational achievement, and height on a cm produce reliable (Welsch et al., 2021) and valid results (Ped- scale. One participant (with stronger self-reported sexual neault et al., 2021; Schmidt et al., 2017). attraction to adults than to children or adolescents) reported a height of 78 cm, which we corrected to the more likely Self‑Reported Sexual Interests A six-item scale introduced by value of 178 cm. Previous convictions for child sexual abuse, Jahnke and Malón (2019) was administered to assess pedo- rape, or possession of child sexual exploitation material were philic, hebephilic, and teleiophilic interests by collecting attrac- assessed on a binary scale (yes/no) with three separate items tion ratings on a Likert scale ranging from 1 (no sexual interest) (“I have been convicted for child sexual abuse,” “I have been to 10 (maximum sexual interest) with regard to children of each convicted for rape,” “I have been convicted for child pornog- sex before puberty, adolescents of each sex in early stages of raphy offenses”). puberty, and sexually mature adult men and women. Categorizing Participants’ Sexual Interests Adapted Edinburgh Handedness Inventory We used the and Oending Sta ff tus modified version of the Edinburgh Handedness Inventory (EHI, Dyshniku et al., 2015). Participants had to decide Classification Based on Self‑Reported Sexual Attraction whether they preferred the left or right hand for presented activities such as writing or drawing (forced choice). Due to In the pedophilic subsample, we retained only men who a technical mistake, it was not possible to choose both the left reported more sexual attraction to prepubescent children than and the right hand in cases where participants did not have to early-to-mid-pubertal adolescents or adults (by subtracting a clear preference (ambidexterity, note that this mistake was the maximum reported sexual attraction to either prepubertal fixed in Study 2 ). Like Dyshniku et al. (2015), we removed girls or boys from the maximum self-reported sexual attrac- the items “using a broom” and “opening a box” because of tion to either of the four other categories [pubescent girls and poor factor analytic loadings. Based on the responses to the boys, sexually mature men and women]). The teleiophilic eight remaining items (α = 0.96), we computed a laterality subsample consisted solely of men indicating highest sexual quotient by dividing the difference score of left and right attraction to adults (subtracting the maximum sexual attrac- responses through the sum of left and right responses. Higher tion to either sexually mature men or women from the maxi- scores indicate a stronger preference for the right hand. mum sexual attraction to either of the four remaining stimuli classes). Those whose sexual attraction was highest for early- to-mid-pubertal adolescents or equally high for pubescent and prepubescent stimuli were categorized as the hebephilic Footnote 1 (continued) subsample. Hence, the hebephilic group included seven indi- the working memory test, which raised doubts whether the digit span viduals with an equally strong attraction to prepubescent and trials were performed according to instructions. 1 3 854 Archives of Sexual Behavior (2022) 51:849–866 pubescent children. For the present analyses, the pedophilic and hebephilic group were combined into a pedohebephilic subsample. As categorization was based entirely on compari- sons of self-reported sexual interests to different groups of people, a difference of at least one unit between the stimulus categories on the sexual interest scale was sufficient for cate- gorization. Yet, one person in the teleiophilic group indicated a sexual interest in prepubescent children that surpassed the midpoint of the 10-point scale. Data from nine men who reported to have equally strong teleiophilic attraction and sexual attraction to one of the remaining four categories were excluded from the analyses. We collected data from 89 self-reported teleiophilic men and 101 self-reported pedohebephilic men. Table 2 gives an overview of self-reported maximum sexual attraction ratings as a function of sexual interest groups and sexual maturity status. The pedohebephilic group was then split up into two subgroups according to their self-reported status regarding previous convictions for sexual offending (child pornography offending and/or child sexual abuse, see Table  2 for more infor- mation on the nature of the convictions). In the teleiophilic group, only one person self-reported a sexual offense involv - ing children and was discarded from the analyses. The groups will be referred to as Pedo-SO (pedohebephilic men who do not report sexual offending), Pedo + SO (pedohebephilic men who do report sexual offending), and Tel-SO (teleiophilic men who do not report sexual offending) in results section. Hence, analyses based on sexual attraction and sexual offending status included 71 Pedo-SO, 30 Pedo + SO, and 88 Tel-SO. Classification Based on Viewing Time For 42 participants, we could not record any response laten- cies likely due to their use of anonymity software which did not permit the task to run properly. Response latency outliers were screened for each single trial utilizing the adjbox func- tion from the R package robustbase (Maechler et al., 2019). This procedure is based on a robust nonparametric modifi- cation of the standard Tukey criterion that specifically fits skewed distributions (Hubert & Vandervieren, 2008) which are notorious for response latency data. Empirically deter- mined outliers were set to missing values (6% of the trials). We calculated average response latencies for each Tanner by stimulus sex category, if we had recorded at least one value for each of the relevant stimulus categories (no cases were deleted because of outlier exclusion). We categorized groups based on the difference score between maximum response latencies to either male or female Tanner stage 5 and Tanner stage 1 stimuli. Similar difference coefficients have been shown to produce the most valid differentiation between teleiophilic and pedohebephilic groups (Schmidt et al., 2017). Participants with positive difference scores were classified as pedohebe- philic subsample (although technically reflecting pedophilic 1 3 Table 2 Self-reported maximum sexual attraction to female and male members of each category of sexual maturity, viewing time scores, and sexual offending status among the pedophilic, hebephilic, and teleiophilic group (Study 1, Study 2) Study 1 Study 2 a a Variable Pedophilia Hebephilia Teleiophilia Pedophilia Hebephilia Teleiophilia M (SD) n M (SD) n M (SD) M (SD) M (SD) n M (SD) n M (SD) n Self-reported attraction to prepubescents 9.57 (0.64) 53 6.00 (2.83) 48 1.35 (0.97) 89 9.74 (0.87) 137 6.77 (2.43) 141 1.30 (1.09) 317 Self-reported attraction to early–mid-pubescents 6.28 (2.05) 53 9.50 (0.83) 48 2.60 (1.95) 89 6.31 (2.31) 137 9.62 (0.76) 141 1.78 (1.71) 317 Self-reported attraction to adults 3.42 (2.48) 53 4.38 (2.45) 48 9.54 (1.02) 80 3.69 (2.46) 137 5.18 (2.54) 141 9.62 (0.91) 317 Viewing time score T1—T5 916 (970) 31 395 (769) 35 -685 (816) 84 - - - - - - Viewing time score T1,2,3- T4,5 - - - - - - 355 (869) 91 162 (781) 90 -831 (902) 313 Prior convictions for sexual offenses combined (in %) 26.4 53 33.3 48 1.1 89 14.6 137 13.5 141 1.3 317 Prior convictions for child sexual abuse (in %) 7.5 53 16.7 48 0.0 89 6.6 137 7.1 141 0.9 317 Prior convictions for rape (in %) 0.0 53 0.0 48 0.0 89 2.2 137 0.7 141 0.0 317 Prior convictions for child pornography offenses (in %) 20.8 53 25.0 48 1.1 89 12.4 137 7.8 141 0.3 317 Includes seven (Study 1) and 31 (Study 2) participants with an equally strong sexual attraction to prepubescents and early-to-mid-pubescents Archives of Sexual Behavior (2022) 51:849–866 855 Table 3 Planned comparisons (Helmert contrasts): Self-reported pedohebephilic men vs. teleiophilic men (Study 1) Variable Pedohebephilia (P –SO, Teleiophilia, no sexual (P-SO, P + SO) vs. T-SO P + SO) offending (T-SO) M (SD) n M (SD) n t (df) p d *** b Attraction to prepubescents 7.87 (2.68) 101 1.32 (0.93) 88 − 22.15 (75.86) < .001 -3.20 *** Attraction to early–mid-pubescents 7.81 (2.26) 101 2.53 (1.87) 88 − 17.00 (186) < .001 − 2.54 *** b Attraction to adults 3.87 (2.50) 101 9.55 (1.03) 88 18.76 (65.62) < .001 2.92 *** b Age 37.52 (12.44) 101 32.53 (11.45) 88 − 4.00 (104.31) < .001 − 0.42 *** Viewing time score T1—T5 640 (901) 66 -692 (818) 83 − 9.46 (146) < .001 − 1.57 Height 180.54 (6.55) 101 180.55 (6.52) 88 0.21 (186) .836 0.00 EHI Laterality Index 0.83 (0.48) 101 0.88 (0.43) 88 0.63 (186) .535 0.10 Head injuries before age 13 0.24 (0.67) 101 0.27 (0.64) 88 0.46 (186) .619 0.05 Head injuries after age 13 0.20 (0.69) 101 0.31 (0.81) 87 0.69 (185) .544 0.14 Social desirability 1.97 (0.56) 100 1.86 (0.50) 88 -1.47 (185) .144 − 0.20 *** p < .001 (two-sided) d = M –M / SD , calculated using the cohen.d function of the R package psych 1 2 pooled We used Welch’s correction due to unequal variances (as indicated by Levene test for equality of variances, center = median) p-value based on 1,000 bootstrap samples due to severe deviations from the assumption that residuals are normally distributed preferences for prepubescent children), while participants with hebephilic men showed positive average scores (indicating negative difference scores were coded as teleiophilic subsam- stronger sexual attraction to children at Tanner stage 1 than ple. Based on their VT profiles, we were able to categorize 67 adults at Tanner stage 5), while teleiophilic men achieved participants as pedohebephilic and 90 participants as teleio- negative scores (indicating stronger attraction to adults than philic. When considering offending status, this left us with 46 to children, see Table 2). VT-based Pedo-SO, 21 Pedo + SO, and 86 Tel-SO. Sample Description Statistical Procedure for Main Analyses Participants were rather well-educated (with 72%, 47%, and Reverse Helmert contrasts were employed for planned 82% reporting to have achieved University entrance cer- comparisons to assess whether there were any differences tificates in Germany for Pedo-SO, Pedo + SO, and Tel-SO, between Pedo-SO, Pedo + SO, and Tel-SO. Specifically, we respectively). Pedo + SO were significantly less educated compared Tel-SO with the mean of Pedo-SO/ + SO and Pedo- than Tel-SO (χ = 13.99, df = 1, p = 0.001; ϕ = 0.34), while SO with Pedo + SO. Because we had planned comparisons, we detected no difference between Pedo-SO and Tel-SO we bypassed omnibus analysis of variance. For some analy- (χ = 2.24, df = 1, p = 0.135; ϕ = 0.12). Pedo-SO/ + SO were ses, we opted against the use of parametric tests because the older than Tel-SO, and Pedo + SO were older than Pedo-SO distribution of residuals showed severe deviations from the (both significant, Tables  3, 4). Sample descriptions based on normal distribution or because of unequal variances (hetero- VT-inferred classifications can be found in Supplement A. scedasticity). In these cases, we used appropriate robust tests (bootstrapping or Welch corrections). Group Differences Unless stated otherwise, results reported here refer to classifi- Results cation based on self-report. Readers can access results based on VT-based classification in the Supplemental Materials. Agreement Between Self‑Reported and Viewing We could not detect significant differences for any of the clas- Time‑Inferred Sexual Interests sification procedures (see Tables  3, 4 for self-report-based classic fi ation and Table S1-S2 in Supplement B for VT-based The two classification procedures led to similar results for classification,), which was expected given the low statistical 127/150 (85%) men who were sorted concordantly as either power. Therefore, we put a stronger focus on effect size point teleiophilic or pedohebephilic based on self-report and VT estimates than the significance of the effects. (χ = 70.98, df = 1, φ = 0.69, p < 0.001). Pedophilic and 1 3 856 Archives of Sexual Behavior (2022) 51:849–866 Table 4 Planned comparisons (Helmert contrasts): Self-reported pedohebephilic men with vs. without convictions for sexual offending (Study 1) Variable Pedohebephilia, no sexual Pedohebephilia, sexual P-SO vs. P + SO offending (P –SO) offending (P + SO) M (SD) n M (SD) n t (df) p d Attraction to prepubescents 7.83 (2.77) 71 7.97 (2.48) 30 0.24 (60.60) .809 0.05 Attraction to early–mid-pubescents 7.63 (2.26) 71 8.23 (2.25) 30 1.32 (186) .188 0.27 Attraction to adults 3.92 (2.44) 71 3.77 (2.67) 30 − 0.26 (50.34) .794 − 0.06 *** b Age 34.03 (10.09) 71 45.80 (13.69) 30 4.25 (42.9) < .001 1.06 Viewing time score T1—T5 552 (879) 43 805 (938) 23 1.15 (146) .253 0.29 Height 180.85 (6.54) 71 179.83 (6.62) 30 − 0.71 (186) .479 − 0.16 EHI Laterality Index 0.83 (0.49) 71 0.83 (0.46) 30 − 0.01 (186) .994 0.00 Head injuries before age 13 0.25 (0.73) 71 0.20 (0.48) 30 − 0.38 (186) .652 − 0.08 Head injuries after age 13 0.16 (0.58) 71 0.30 (0.92) 30 0.84 (185) .463 0.20 Social desirability 1.95 (0.55) 71 2.01 (0.60) 29 0.52 (185) .601 0.11 *** p < .001 (two-sided) d = M –M / SD , calculated using the cohen.d function of the R package psych 1 2 pooled We used Welch’s correction due to unequal variances (as indicated by Levene test for equality of variances, center = median) p-value based on 1,000 bootstrap samples due to severe deviations from the assumption that residuals are normally distributed Further data inspection revealed that the Pedo + SO group head injuries, we determined significance based on 1,000 was about 1 cm shorter than the Pedo-SO and Tel-SO (see bootstrap samples because these variables showed severe Table 3). Our results also suggested a higher likelihood for deviations from the assumption of normality. There was no Pedo-SO/ + SO to be left-handed compared to Tel-SO (see significant relationship between age and head injuries before Table 3), while there was no indication for differences in age 13 (Pedo-SO: r = .07, 95%CI = [−.11, .40], Pedo + SO: handedness depending on sexual offending status among r = .20, 95%CI = [−.25, .50], Tel-SO: r = .01, 95%CI = [−.19, the pedohebephilic group (see Table 4). Furthermore, our .22]). We found, however, a significant negative relationship data indicated lower rates of head injuries before age 13 for between age and head injuries after age 13 for non-offending Pedo-SO/ + SO compared to Tel-SO (see Table 3). Again on a teleiophilic participants (Pedo-SO: r = .18, 95%CI = [−.18, descriptive level, Pedo + SO reported more head injuries after .47] Pedo + SO: r = .24, 95%CI = [−.16, 0.47], Tel-SO: age 13 than Pedo-SO (Table 4). All three groups achieved r = −.14, 95%CI = [−.25, -.003]). Similar non-effects were similar scores on the social desirability scale (Tables 3, 4). obtained for the VT-inferred classification, with the excep- In group comparisons based on VT-inferred classifications of tion of a significant negative link between height and age sexual orientation, we achieved results in similar directions, among VT-based Pedo + SO (see Supplement C). Taking with the sole difference that Pedo + SO were about 0.40 cm into account the number of tests, size, and direction of these taller than Pedo-SO, and that Pedo-SO/ + SO appeared more effects, these observations speak against a systematic pattern likely to be right-handed than Tel-SO (all ns., see Tables for the head injuries measure. Nevertheless, they indicate, S1—S2 in Supplement B). albeit not consistently, that differences in age may explain some of the (nonsignificant) differences for the contrasts Control Analyses based on height. As group differences in age might have biased our results, we assessed links between age, height, and head injuries before Discussion and after age 13 separately for each of the six tested subgroups (i.e., [1] self-reported Pedo-SO, Pedo + SO, and Tel-SO; [2] Study 1 showed that it is feasible—at least for a majority of VT-inferred Pedo-SO, Pedo + SO, and Tel-SO). For the self- the sampled individuals—to run an online version of the VT report classification, we found no significant link between measure among self-identified males with sexual interest in age and height (Pedo-SO: r = .08, p = .482, 95%CI = [−.15, children and adolescents. For the first time, these results extend .31], Pedo + SO: r = −.33, p = .072, 95%CI = [−.62, .03], Tel- the validation of VT tasks as measures of pedohebephilic sexual SO: r = −.13, p = .217, 95%CI = [−.33, .08]). Note that for interest (Schmidt et al., 2017) to self-identified pedohebephilic 1 3 Archives of Sexual Behavior (2022) 51:849–866 857 men from the community which can be regarded as the most countries of residences were allowed for participation (task direct test of VT validity hitherto. Nevertheless, results of Study language was English): Australia, Austria, Belgium, Can- 1 are subject to typical caveats associated with research based ada, Czech Republic, Denmark, Estonia, Finland, France, on self-reported data, retrospective questionnaires, and online Germany, Greece, Iceland, Italy, Lithuania, Luxembourg, research (see general discussion for a detailed elaboration). In Netherlands, New Zealand, Norway, Poland, Portugal, Slo- addition, statistical power was insufficient. Yet, descriptive vakia, Spain, Sweden, UK, and the USA. In sum, we recruited analyses indicated that differences between pedohebephilic and 331 participants via B4U-ACT and 318 via MTurk. After non-pedohebephilic groups might be driven by a confound of the exclusion of 17 people who did not disclose their sex or (self-reported) sexual offending status. Furthermore, pedohe- reported to be female or other, this left us with a total of 632 bephilic and teleiophilic men reported similar rates of head participants. Detailed information on how we categorized injuries before age 13, irrespective of their offending status. participants based on their sexual interests and offending Descriptively, our data appear to be more in line with research status is presented in a separate section after the description pointing toward a link between markers for neurodevelopmen- of our measures. tal differences and norm-breaking/criminal behavior, instead of pedohebephilia per se (with the exception of left-handedness, Measures which was more prominent in the pedohebephilic group) calling for a more stringent test based on adequate statistical power. Measurements were administered in the same order as they This was sought to achieve in Study 2. appear in the following section. Additionally, we assessed information about traumatic childhood experiences and non- negative sexual experiences in childhood for a separate pub- Study 2 lication (Jahnke et al., 2021). Method Viewing Time Measure of Sexual Interest We employed a similar VT task as in Study 1, again using stimulus material Participants and Sample Size Calculations from the Not Real People set (Pacific Psychological Assess- ment Cooperation, 2004). Deviating from Study 1, we added To reach more potential participants, which is necessary to pictures of early (Tanner stages 2–3) and late adolescents address the statistical power issue, and to make sure that sam- (Tanner stage 4), alongside pictures of adults (Tanner stage ples in Study 1 and 2 are (at least largely) independent, Study 5) and prepubescent children (Tanner stage 1). To shorten 2 targeted English-speaking men. We conducted sample size test duration, we presented only four pictures per sex and calculations for all markers for neurodevelopmental differ - Tanner stage category. ences based on effect sizes from the literature (see Table  1). We estimated that among all cognitive markers for neurode- Self‑Reported Sexual Interests We used the English version velopmental differences, self-reported height would show of the measure from Study 1. the smallest average difference between pedohebephilic and teleiophilic participants. Hence, we set our sampling goal to International Cognitive Ability Resource Sample Test To 620 participants, based on estimations with G*Power (setting measure cognitive ability, we used the 16-item sample test α = 0.05 and β = 0.20, one-sided testing for height, two-sided from the public-domain International Cognitive Ability testing for all other hypotheses, Faul et al., 2007). The study Resource (ICAR; Condon & Revelle, 2014), which consists of was approved by the institutional review board of the MSH letter and number series, matrix and verbal reasoning items, Medical School. and three-dimensional rotation items. Previous research has The study was advertised as a survey on wanted and established that the ICAR 16-item sample test is a reliable unwanted childhood sexual experiences, cognitive develop- and valid measure of global cognitive ability (Condon & ment, and sexual interests in children or adults among men Revelle, 2015; Merz et al., 2020). The (self-administered) from the community. We collected data via the B4U-ACT ICAR-16 shows high convergent validity with the Wechsler support group for people with pedohebephilia between July Adult Intelligence Scale (fourth edition) administered by 2018 and March 2019, offering to donate 1.50$ for each a trained clinician (r = 0.81 and r = 0.94 with the manifest participant to B4U-ACT (with a maximum donation sum and the latent score of the WAIS-IV, respectively, Young & of $300). We furthermore recruited participants from the Keith, 2020). As pedohebephilic participants might have a MTurk workforce. Only male MTurk workers who had been stronger motivation than teleiophilic MTurkers to excel on approved for 100 to 5000 human intelligence tasks with an the IQ test, it was planned to compare their responses with overall approval rate of 80% or more were eligible for the scores obtained among Internet users who wanted to test their study and received a payment of two US$. The following cognitive abilities in a previous study (Condon & Revelle, 1 3 858 Archives of Sexual Behavior (2022) 51:849–866 2015). To match procedures employed in this comparison men reported sexual interests surpassing the midpoint of the study, we presented the test without time restraints. For 18 respective scales. Group means for sexual attraction ratings participants (all recruited via B4U-ACT) we could not cal- are displayed in Table 2. In the teleiophilic group, four par- culate sum scales due to missing values. Internal consistency ticipants self-reported a sexual offense and were therefore was good (Kruder-Richardson 20 = 0.84). discarded from the T-SO group. Analyses based on sexual attraction and sexual offending status included 239 Pedo-SO, Adapted Edinburgh Handedness Inventory We used the 39 Pedo + SO, and 313 Tel-SO. English version of the EHI (Dyshniku et al., 2015) employed in Study 1, again without the broom and box items (α = 0.89, Classification Based on Viewing Time For 110 participants, no missing values). we could not record any response latency data, most likely due to the use of anonymity software. Using the same method Head Injuries As in Study 1, we used items from Blanchard to identify outliers as in Study 1, 4% of recorded trials were et al. (2013) to assess head injuries involving unconscious- marked as outliers. All participants had to have at least one ness before and after age 13. One participant recruited via valid response for each Tanner stage by stimulus sex category B4U-ACT left these questions unanswered. (note that this requirement was fulfilled in all cases where VT scores were recorded). Viewing time data were analyzed in Sociodemographic Information (Including Height and Previ‑ the same way as described in Study 1, with one exception: ous Convictions) As in Study 1, we assessed sex, age, and As Study 2 included pictures from Tanner stages 1 to 5, we educational achievement (i.e., having or not having obtained additionally calculated difference scores between maximum a Bachelor's degree or higher). Participants were allowed to average response latencies to either male and female stimuli choose whether to report their height in the metric (cm) or separately for each of the five Tanner stages. To determine the imperial system (the latter were subsequently transferred pedohebephilia, we subtracted maximum average reaction to cm). For height, we corrected obvious typing mistakes times to either male or female stimuli in Tanner stages 4 and 5 in four cases (e.g., 88 inches were changed to 8 inches). To (late pubescence and adulthood) from Tanner stages 1, 2, and determine outliers, we estimated fences based on the Tukey 3 (pre-peri-pubescence). Participants with positive difference criterion (1.5 interquartile ranges). This led to the exclusion scores were classified as pedohebephilic, while participants of five data points above 198.12 cm and six below 157.48 cm. with negative difference scores were classified as teleiophilic. Cases with heights outside the fences were about equally Thus, we categorized 60 participants as pedophilic (i.e., hav- split between participants from MTurk (2 above, 3 below) ing higher scores for Tanner stage 1 than for Tanner stages 2 and B4U-ACT (3 above, 3 below). Twelve participants from to 5), 119 as hebephilic (i.e., having higher values for Tanner B4U-ACT and two from MTurk did not report their height. stages 2 and 3 than for Tanner stages 1, 4, and 5), and 326 as For the assessment of previous convictions, we administered teleiophilic (i.e., having higher values for Tanner stages 4 and the same items as in Study 1. 5 than for Tanner stages 1 to 3). When accounting for offend- ing status previous convictions for sexual offending (child Categorizing Participants’ Sexual Interests and Oending ff pornography offending, rape, and/or child sexual abuse, see Status Table 2 for further information), 154 were categorized as Pedo-SO, 25 as Pedo + SO, and 326 as Tel-SO. Classification Based on Self‑Reported Sexual Attraction We formed groups based on self-reported sexual interests fol- lowing procedures in Study 1. This led us to categorize 317 participants as teleiophilic men and 278 as pedohebephilic Results men (141 with a predominant pedophilic interest). Twenty- one participants from the self-reported teleiophilic group Agreement Between Self‑Reported and Viewing were recruited via B4U-ACT, while seven participants Time‑Inferred Sexual Interests from the self-reported pedohebephilic group were sampled on MTurk. Those participants were not excluded from the For the 493 participants that we could classify based on self- analyses. Thirty-seven participants who reported equal reported and VT-inferred sexual interests, classification pro- maximum sexual attraction to adult and prepubertal/early- cedures led to similar results (χ = 123.32, df = 1, φ = 0.50, to-mid-pubertal persons were excluded. Among self-reported p < 0.001). Of the self-reported teleiophilic participants, 267 teleiophilic participants, only seven (2%) reported a sexual (85%) showed a teleiophilic pattern in the VT task, while 114 interest in prepubescent girls or boys surpassing the mid- (63%) of self-reported pedohebephilic participants showed point of the sexual attraction scale. Regarding girls or boys in VT results indicating pedohebephilic interests. The pedo- early-to-mid-puberty, only 20 (6%) self-reported teleiophilic philic and hebephilic groups achieved on average positive 1 3 Archives of Sexual Behavior (2022) 51:849–866 859 Table 5 Planned comparisons (Helmert contrasts): Self-reported pedohebephilic men vs. teleiophilic men (Study 2) Variable Pedohebephilia (P –SO, Teleiophilia, no sexual (P-SO, P + SO) vs. T-SO P + SO) offending (T-SO) M (SD) n M (SD) n t (df) p d *** b Attraction to prepubescents 8.23 (2.36) 278 1.29 (1.09) 313 − 38.97 (75.36) < .001 − 3.87 *** b Attraction to early–mid-pubescents 7.99 (2.38) 278 1.77 (1.71) 313 − 30.31 (92.57) < .001 − 3.04 *** b Attraction to adults 4.44 (2.61) 278 9.64 (0.91) 313 23.15 (53.09) < .001 2.73 Age 34.44 (12.98) 273 35.15 (11.06) 313 − 1.21 (113.35) .228 0.06 *** Viewing time score T1,2,3—T4,5 259 (830) 181 − 839 (904) 309 − 10.77 (487) < .001 − 1.25 Height 178.89 (7.55) 259 177.66 (7.18) 307 0.11 (563) .545 − 0.17 *** d EHI Laterality Index 0.68 (0.61) 278 0.68 (0.58) 313 − 0.77 (588) .387 0.00 ** b ICAR 9.86 (4.18) 260 7.24 (3.52) 313 − 3.23 (78.43) .002 − 0.69 Head injuries before age 13 0.25 (0.66) 277 0.25 (0.68) 313 − 0.20 (587) .867 0.00 b d Head injuries after age 13 0.20 (0.64) 276 0.30 (0.78) 313 0.01 (64.35) .989 0.14 *** ** p < .001, p < .01 (two-sided) d = M –M / SD , calculated using the cohen.d function of the R package psych 1 2 pooled We used Welch’s correction due to unequal variances (as indicated by Levene test for equality of variances, center = median) One-sided p p-value based on 1,000 bootstrap samples due to severe deviations from the assumption that residuals are normally distributed VT mean scores (indicating a stronger attraction to pedohe- intelligent than P + SO (Table 6). Furthermore, unobtrusively bephilic stimuli on Tanner stage 1–3 than to late/postpubes- recorded response latencies during the IQ test revealed that cent stimuli on Tanner stage 4 or 5), while the self-reported the Pedo-SO group had an almost twice as long average teleiophilic group showed negative VT scores (indicating a completion time as the Pedo + SO and the Tel-SO group stronger relative attraction to physically mature adults, see (with median scores of 15.09 min compared to 7.65 min and Table 2). 8.40 min, based on reaction time data from 153 Pedo-SO, 29 Pedo + SO, and 310 Tel-SO). To compare results with a sample that was arguably more motivated to score high, we Sample Description used data from the ICAR project (Condon & Revelle, 2015) to extract means and standard deviations for each of the items Educational levels were relatively high as 53%, 49%, and from the ICAR 16-item sample test among men over 18 years 62% indicated to have achieved an associate degree, BA (n between 4220 and 13,216, average n = 9324, note that the degree, or higher among Pedo-SO, Pedo + SO, and Tel-SO, ICAR items were delivered untimed in both our survey and respectively. We detected no significant difference between the ICAR project). All of these participants had solved a var- Tel-SO, Pedo-SO, and Pedo + SO regarding educational level ying number of the ICAR item set online in order to receive (χ = 5.27, df = 2, φ = 0.09, p = 0.072). Sexual offenses were a personalized feedback (Condon & Revelle, 2014). These reported mostly by pedohebephilic participants (Table 2). individual means and standard deviations were added to cre- Age did not differ significantly between Pedo-SO/ + SO and ate average sum scores (M = 8.60, SD = 7.42). t-tests with Tel-SO, but Pedo + SO were significantly older than Pedo-SO Welch-adjusted degrees of freedom based on these summary (Tables 5–6). Sample descriptions based on VT are presented statistics revealed that Pedo-SO/ + SO (M = 9.86, SD = 4.18, in Supplementary Material A. n = 260) scored higher than participants from the ICAR project. Yet, albeit significant ( t(306.45) = 4.66, p < 0.001, d = 0.21), effect sizes were smaller than the ones we retrieved Group Differences for the comparison with the Tel-SO within the current study. For the self-report-based classification, we found no dif- Pedo-SO/ + SO had higher scores on the ICAR test than Tel-SO, even ferences between Pedo-SO/ + SO and Tel-SO regarding when controlling for completion time via an ANCOVA design (F(1, 489) = 25.25, p < .001). Yet, note that although ANCOVAs represent a height, laterality index, and head injuries before and after common strategy to "control" for confounds, we have doubts that this age 13 (Table 5). Yet, we detected significant differences in strategy is appropriate and leads to meaningful conclusions (see Miller IQ scores, with Pedo-SO/ + SO achieving higher scores than & Chapman, 2001 for a non-technical discussion of common misuses Tel-SO. P-SO also emerged as significantly taller and more of ANCOVA in psychopathology research). 1 3 860 Archives of Sexual Behavior (2022) 51:849–866 Table 6 Planned comparisons (Helmert contrasts): Self-reported pedohebephilic men with vs. without convictions for sexual offending (Study 2) Variable Pedohebephilia, no sexual Pedohebephilia, sexual P-SO vs. P + SO offending (P –SO) offending (P + SO) M (SD) n M (SD) n t (df) p d Attraction to prepubescents 8.16 (2.41) 239 8.69 (1.92) 39 1.55 (59.48) .128 0.23 Attraction to early–mid-pubescents 7.98 (2.44) 239 8.03 (2.05) 39 0.12 (57.13) .908 0.02 ** b Attraction to adults 4.63 (2.53) 239 3.31 (2.81) 39 − 2.76 (48.56) .008 − 0.52 ** b Age 33.59 (13.14) 234 39.51 (10.84) 39 3.06 (58.31) .003 0.46 Viewing time score T1,2,3—T4,5 258 (860) 152 262 (659) 29 0.02 (487) .982 0.00 ** Height 179.38 (7.34) 224 175.77 (8.18) 35 − 2.72 (563) .007 − 0.49 EHI Laterality Index 0.66 (0.62) 239 0.79 (0.54) 39 1.27 (588) .167 0.21 *** b ICAR 10.35 (4.00) 224 6.78 (4.02) 36 − 4.95 (46.81) < .001 − 0.90 Head injuries before age 13 0.24 (0.61) 239 0.29 (0.90) 38 0.40 (587) .763 0.07 b c Head injuries after age 13 0.16 (0.56) 239 0.43 (0.99) 37 1.64 (39.63) .131 0.44 *** ** p < .001, p < .01 (two-sided) d = M –M / SD , calculated using the cohen.d function of the R package psych 1 2 pooled We used Welch’s correction due to unequal variances (as indicated by Levene test for equality of variances, center = median) p-value based on 1,000 bootstrap samples due to severe deviations from the assumption that residuals are normally distributed When the classification of sexual attraction was based on General Discussion VT, none of the contrasts for markers for neurodevelopmental differences reached significance (see Supplementary Mate - We could not corroborate our first hypothesis that pedo- rial Tables S3–S4). Note that sample sizes were smaller in hebephilic men from our two community samples differed the alternative classification, so failure to reach significance significantly from teleiophilic men with respect to height, was more likely. Yet, differences descriptively tended to be intelligence, non-right-handedness, or rate of head inju- similar in magnitude and direction, with the exception of ries during childhood. We did, however, find evidence for Pedo-SO/ + SO showing a stronger preference for the right the second hypothesis that pedohebephilic men who have hand compared to Tel-SO. sexually offended were smaller and less intelligent than non- offending pedohebephilic men. Replicating Gerwinn et al.'s Control Analyses (2019) findings, the current studies do not yield support for the theory that sexual interests in children among community To determine whether age could be a confound, we assessed men are linked to neurodevelopmental differences. Yet, in its link to IQ, height, and head injuries before and after age contrast to Gerwinn et al. (2019), which suffered from lacking 13 separately for each of the six tested subgroups (i.e., [1] statistical power, Study 2 was able to test differences between self-reported Pedo-SO, Pedo + SO, and Tel-SO; [2] VT- pedohebephilic and teleiophilic men at least 1-β = 0.80, with inferred Pedo-SO, Pedo + SO, and Tel-SO). For variables the exception of height. that deviated severely from the assumption of normality (in Because of the limitations of the present case–control this case head injuries before or after 13), we used boot- design and the impossibility to prove a null hypothesis in strapping procedures. For the self-report based classifica- classical hypothesis testing, our findings cannot conclusively tion, only 2/12 tests showed a significant link between older disprove the idea of a general link between indicators of neu- age and any of the dependent variables (head injuries before rodevelopmental perturbations and pedohebephilia. Never- age 13 and ICAR among Pedo-SO, r = 0.13, 95%CI based theless, the present data are positively commensurate with on bootstrap samples = [0.008, 0.29] and r = 0.14, p = 0.033, the notion that indicators of neurodevelopmental perturba- 95%CI = [0.01, 0.27], respectively). For the VT-based clas- tions such as, particularly, height and intelligence are linked sification, we could also only detect significant results for to (self-reported) offending status when pedophilic interest 1/12 tests (see Supplemental Material C). Considering the is kept constant in group comparisons. This dovetails with large number of tests and the direction of the effects, the the fact that all markers of neurodevelopmental perturba- evidence for age confounds was weak, as the number of sig- tions as assessed here are empirical correlates of criminal nificant tests does not fall far from the margin of what would behavior—as outlined in introduction (see also Beckley et al., be expected by chance at the given significance threshold of 2014; Bogaert, 2001; McKinlay et al., 2014; Moffitt et al., p < 0.05. 1994). Future research is needed to clarify if the importance 1 3 Archives of Sexual Behavior (2022) 51:849–866 861 of neurobiological alterations might have been overstated in self-referred pedohebephilic patients in treatment projects the previous literature, as part of the variance attributed to (Jahnke et al., 2015). pedohebephilia may in fact have to be attributed to other fac- The generalizability of the present findings is further con- tors due to the lack of specificity of the neurodevelopmental strained by the WEIRDness (Western, Educated, Industrial- perturbations hypothesis. ized, Rich, Democratic; Henrich et al., 2010) of the present samples. Note also that we cannot guarantee that Study 1 and Limitations and Outlook 2, without exception draw from two completely independent participant pools. It is possible that some participants are flu- Our results are subject to a number of caveats, some of which ent in both German and English, have visited both German are specific to our setting and methodology, while others and English language web forums for pedohebephilic people, are generally associated with research on proxy measures of and have participated in both studies. Future research draw- neurodevelopmental perturbations. First and foremost, cor- ing on online samples of pedohebephilic men should include relational analyses are unfit to prove causation—a fact that an item to identify and potentially screen out participants applies to most studies in the field of the neurodevelopmental from previous studies on the neurodevelopmental markers perturbations hypothesis and that increases the risk of fall- of pedohebephilia. ing victim to third variable confounds in case–control study Furthermore, both studies only included relatively small designs. numbers of pedohebephilic men who reported convictions for Additionally, our study populations are neither representa- sexual offenses. This means that even though some contrasts tive for teleiophilic nor pedohebephilic men. While clini- yielded significant results, small changes in participant char - cal, institutional, or forensic samples are subject to selection acteristics can lead to fluctuations in the results. Although biases as well, this means that results have to be interpreted difficult to attain, replications with larger samples of men carefully with respect to confounds and (self-selection) with sexual interests in children who have sexually offended biases. It is, for instance, possible that tall pedohebephilic are needed to corroborate these effects with more certainty. men with high levels of cognitive functioning were more Preferably, these should be men from the community to con- likely to participate than those with lower height or lower trol for selection factors related to being institutionalized in levels of cognitive abilities, given that they were aware of forensic settings. hypotheses from the academic literature. Yet, as online stud- Additionally, readers need to keep in mind that offending ies generally tend to oversample younger and more educated status was determined based on self-reported convictions for participants, this limitation also applies to the teleiophilic rape, child sexual abuse, and/or child pornography offenses groups. It is also possible that links between markers of neu- alone. Hence, there may be people in our non-offending rodevelopmental perturbations and pedohebephilia exist in sample who have sexually offended but were not detected or the population but cannot be identified when assessing sub- convicted, which makes it impossible to tell whether effects groups with high computer literacy. Even if pedohebephilia are a function of criminal sexual behavior or of being con- was linked to such markers, it may be impossible to detect victed for such acts. Previous research also indicates that men these links in samples without individuals with low levels convicted for child pornography offenses may differ from of cognitive functioning. Yet, note that in forensic settings, men convicted for child sexual offending on a number of pedohebephilic men who have been convicted for child por- relevant variables, such as antisociality (Babchishin et al., nography offending (i.e., indicating at least average levels 2015). Hence, it is possible that we would have found larger of computer literacy and cognitive capacity) showed more differences between the groups with and without a history markers of neurodevelopmental perturbations than teleio- of sexual offending, if we had had a higher rate of pedohebe- philic men who have committed the same crime (Blanchard philic men convicted for child sexual offending compared to et al., 2007). In the absence of representative samples of tel- child pornography offending. Yet, note that a study compar - eiophilic and pedophilic men, it is impossible to determine ing pedohebephilic people with either of these convictions if our results are due to such ceiling effects. shows more similarities than differences between the groups Moreover, participants from the pedohebephilia groups (Neutze et al., 2011). Furthermore, because we did not assess were recruited via forums addressing sexual attraction to chil- previous non-sexual convictions, we could not screen out dren, where they may have been looking for support because people with a significant non-sexual offense history as an they experienced increased levels of distress. Indeed, previ- indicator of antisociality. This is a potential limitation, as ous research indicates that pedohebephilic participants in antisociality may be linked to neurodevelopmental pertur- online surveys generally report higher levels of distress than bations. Yet, given the nature of our surveys and previous population-based or other non-clinical control samples. Yet, research among MTurk samples, our samples are unlikely to it is also shown that they are considerably less distressed than include people with marked antisocial traits. 1 3 862 Archives of Sexual Behavior (2022) 51:849–866 Another concern is that pedohebephilic participants might in its entirety, as the evidence for this hypothesis goes beyond have been more motivated to perform optimally in order to the factors examined here. better the image of their group. These motivational differ - Furthermore, with exception of the VT measure and the ences became apparent in the intelligence measure in Study 2, IQ test, the online environment forced us to rely completely which involved complex puzzles that required eo ff rt to solve. on self-report, which puts restraints on the validity of the data Teleiophilic MTurk workers, who purportedly had less to assessed. The tendency of men to over-report their height is gain from putting extra efforts into the cognitive tests, spent well-documented (Gorber et al., 2007), and pedophilic and much less time on solving the items than the people from teleiophilic men who have committed sexual offenses are no the pedohebephilic group. To get a perhaps more accurate exception to this rule: Fazio et al. (2014) report differences estimation of pedohebephilic men's cognitive abilities, we between actual and self-reported height varying between additionally compared responses from our pedohebephilic 2 and 4 cm, similar to differences observed in non-clinical sample in Study 2 with test data from Condon and Revelle's samples (e.g., Palta et al., 1982). This means that the data we (2014) validity study on the ICAR items. As participants in obtained on height in both our pedophilic and teleiophilic their dataset had been recruited online with the prospect to samples are likely to be biased. However, this is potentially a receive customized feedback, we expected that these partici- common limitation, at least for earlier studies, which, aston- pants had put more effort into solving the presented logical ishingly, either report to have relied on self-reported height puzzles compared to our teleiophilic sample. Yet, while these (e.g., Cantor et al., 2007) or had no access to information on participants scored higher on the test items than teleiophilic how height was assessed (Levenson & Ackerman, 2017; but men from Study 2, pedohebephilic men still emerged as more note that this is less likely to be the fact for newer research, intelligent. Fazio et al., 2017). Despite those problems, online research One of the main drawbacks of online research is the lack may represent the only feasible way of reaching participants of control that researchers can exert on the setting. In con- with pedohebephilia (as a dominant sexual attraction) outside trast to laboratory surveys, we were not able to make sure of clinical or forensic contexts. that participants stayed focused during the IQ test or cor- The present paper only focused on how men with pedo- rectly followed all instructions, and doubts remain whether hebephilic (who have and have not sexually offended) and IQ tests in non-controlled online environments compare teleiophilic interests differed in terms of neurodevelopmental with clinician administered IQ inventories. Yet, note that indicators. Future studies based on larger samples than the previous research found no structural differences regarding present one should consider conducting additional more fine- matrices intelligence scores obtained in online and offline grained analyses (e.g., to differentiate between pedophilic, samples (Ihme et al., 2009). Notably, although the presence hebephilic, and teleiophilic participants or between different of an interviewer can be advantageous, it may also introduce types of sexual offenses involving children). It also needs an expectancy bias on the interviewer’s side. For instance, to be stressed that, following conventions in the field, the clinical staff may look harder for anomalies when examining present study was conducted with a type II error rate of 0.20, pedophilic men, or they may more readily accept a teleio- meaning that there is a 20% chance that the null hypothesis philic man's self-report to have never had an accident leading (i.e., that there is no difference between pedohebephilic and to unconsciousness, prompting them to ask fewer follow-up teleiophilic men) is falsely accepted. questions. Reporting biases due to perceived demand effects on the interviewee’s side (and/or the larger clinical-forensic Conclusions context) are also conceivable. In the present studies, where pedophilic and non-pedophilic men were prompted with the Our findings underscore a general conclusiveness problem same set of questions, such types of bias are unlikely. arising from case–control designs: The interpretation of find- We were also limited in the range of variables we were ings is largely dependent on the composition of the comparison able to assess, as many potentially interesting outcome vari- groups. While there is no such thing as a representative sample ables are difficult (e.g., minor physical anomalies or 2D:4D of men with pedohebephilic interests, studies from community digit ratio; Jordan et al., 2020) or even impossible (e.g., brain settings represent an important corrective to data from pedo- morphology and functioning; Cantor et al., 2008; Massau hebephilic men "who are available for study because they are et al., 2017; Schiffer et al., 2017) to measure in an online either distressed by their sexual interests […] or criminally setting. For obvious reasons, this precludes the assessment charged for sexual offenses" (Seto, 2004, p. 323). Whatever of large numbers of pedohebephilic men without a history of sexual offending as a mostly hidden and hard-to-reach population. On the one hand, this limits our ability to judge The authors are willing to share the results of such more detailed the validity of the neurodevelopmental perturbations theory analyses based on our datasets (e.g., for the purpose of meta-analysis) upon request. 1 3 Archives of Sexual Behavior (2022) 51:849–866 863 processes may ultimately be involved in the development of Declarations pedohebephilic interests, evidence from recent surveys involv- Conflict of interest The authors declare that they have no conflict of ing non-offending pedohebephilic individuals indicates that interest. the claim that pedohebephilia is based on neurodevelopmen- tal perturbations might not be as inevitable and generalizable Consent to Participate Informed consent was obtained from all indi- as previously thought by bearing the potential to disentangle vidual participants included in the study. “institutional caseness,” criminality, and sexual interest (Ger- Consent to Publish Participants consented to the publication of their winn et al., 2018). Moreover, as the neurodevelopmental pertur- data in aggregated form. bation account is silent on the actual processes that are involved in developing pedohebephilia (and which thus have never been Ethical Approval This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the institutional subjected to empirical testing) only proxy measures that itself review board of the MSH Medical School for Study 1 and the institu- are just correlates of the supposed developmental trajectory tional review board of the Technische Universität Dresden for Study 2. have hitherto been examined. The potential of such correlates (i.e., the perturbation proxy measures used here) of correlates Open Access This article is licensed under a Creative Commons Attri- (i.e., one process among possibly many etiological pathways) to bution 4.0 International License, which permits use, sharing, adapta- be confounded with third variables is high and leads to substan- tion, distribution and reproduction in any medium or format, as long tial methodological problems in gathering empirical support as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes for the theory—specifically if it refers to a phenomenon that is were made. The images or other third party material in this article are linked to criminal behavior, publicly despised (Lehmann et al., included in the article's Creative Commons licence, unless indicated 2020), and has a low base-rate (Bártová et al., 2021; Dombert otherwise in a credit line to the material. If material is not included in et al., 2016). the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will Last but not least, it is noteworthy that data from two need to obtain permission directly from the copyright holder. To view a samples corroborated the validity of VT assessments in copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . self-identified community males with pedohebephilic sexual interests by yielding theoretically meaningful die ff rences in a known-groups approach. This can be regarded as the hitherto References most accurate estimator of the validity of VT measures, as prior studies rested on various group comparisons with men Abé, C., Adebahr, R., Liberg, B., Mannfolk, C., Lebedev, A., Eriks- convicted for child sexual abuse. The latter groups, however, son, J., Långström, N., & Rahm, C. (2021). Brain structure and clinical profile point to neurodevelopmental factors involved in are to a large degree only a pedohebephilia proxy as they are pedophilic disorder. Acta Psychiatrica Scandinavica, 143(4), composed of pedohebephilic and teleiophilic individuals thus 363–374. https:// doi. org/ 10. 1111/ acps. 13273 yielding only a conservative (i.e., lower-bound) estimator of Babchishin, K. M., Hanson, R. K., & VanZuylen, H. (2015). Online the magnitude of possible group differences (Schmidt et al., child pornography offenders are different: A meta-analysis of the characteristics of online and offline sex offenders against 2017). children. Archives of Sexual Behavior, 44(1), 45–66. https://doi. org/ 10. 1007/ s10508- 014- 0270-x Supplementary Information The online version contains supplemen- Bailey, A. A., & Hurd, P. L. (2005). Finger length ratio (2D:4D) cor- tary material available at https://doi. or g/10. 1007/ s10508- 021- 02228-w . relates with physical aggression in men but not in women. Bio- logical Psychology, 68(3), 215–222. https:// doi. org/ 10. 1016/j. Acknowledgements We are grateful to David M. Condon for sharing biops ycho. 2004. 05. 001 his expertise on online intelligence assessment and to Ian V. McPhail Bártová, K., Androvičová, R., Krejčová, L., Weiss, P., & Klapilová, for commenting on an earlier draft of this research. K. (2021). The prevalence of paraphilic interests in the Czech population: Preference, arousal, the use of pornography, fantasy, Author Contributions All authors developed the study concept and and behavior. Journal of Sex Research, 58, 86–96. https:// doi. design for Study 1. Data collection for Study 1 was performed by A.K. org/ 10. 1080/ 00224 499. 2019. 17074 68 under the supervision of J.H. and S.J. Data analysis for Study 1 was Becerra García, J. A. (2009). Etiology of pedophilia from a neurode- conducted by S.J. and A.S. S.J. and A.S. developed the study concept velopmental perspective: Markers and brain alterations. Revista and collected and analyzed the data for Study 2. S.J. drafted the paper, de Psiquiatría y Salud Mental, 2(4), 190–196. https:// doi. org/ and A.S. and J.H. provided critical revisions. All authors approved the 10. 1016/ S2173- 5050(09) 70051-2 final version of the paper for submission. Beckley, A. L., Kuja-Halkola, R., Lundholm, L., Långström, N., & Frisell, T. (2014). Association of height and violent criminality: Results from a Swedish total population study. International Funding Open access funding provided by University of Bergen Journal of Epidemiology, 43(3), 835–842. h t t p s : / / d o i . o r g / 1 0 . (incl Haukeland University Hospital). This study received no specific 1093/ ije/ dyt274 funding. 1 3 864 Archives of Sexual Behavior (2022) 51:849–866 Blanchard, R., Christensen, B. K., Strong, S. M., Cantor, J. M., sexual interest in prepubescent children? Journal of Sex Research, Kuban, M. E., Klassen, P., Dickey, R., & Blak, T. (2002). Ret- 53(2), 214–223. https://doi. or g/10. 1080/ 00224 499. 2015. 10201 08 rospective self-reports of childhood accidents causing uncon- Dyshniku, F., Murray, M. E., Fazio, R. L., Lykins, A. D., & Cantor, J. sciousness in phallometrically diagnosed pedophiles. Archives M. (2015). Minor physical anomalies as a window into the pre- of Sexual Behavior, 31(6), 511–526. https:// doi. org/ 10. 1023/A: natal origins of pedophilia. Archives of Sexual Behavior, 44(8), 10206 59331 965 2151–2159. https:// doi. org/ 10. 1007/ s10508- 015- 0564-7 Blanchard, R., Kolla, N. J., Cantor, J. M., Klassen, P. E., Dickey, R., Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: Kuban, M. E., & Blak, T. (2007). IQ, handedness, and pedophilia A flexible statistical power analysis program for the social, behav - in adult male patients stratified by referral source. Sexual Abuse: ioral, and biomedical sciences. Behavior Research Methods, 39(2), A Journal of Research and Treatment, 19(3), 285–309. https://doi. 175–191. org/ 10. 1007/ s11194- 007- 9049-0 Fazio, R. L. (2018). Toward a neurodevelopmental understanding of Blanchard, R., Kuban, M. E., Klassen, P., Dickey, R., Christensen, B. K., pedophilia. Journal of Sexual Medicine, 15(9), 1205–1207. https:// Cantor, J. M., & Blak, T. (2003). Self-reported head injuries before doi. org/ 10. 1016/j. jsxm. 2018. 04. 631 and after age 13 in pedophilic and nonpedophilic men referred for Fazio, R. L., Dyshniku, F., Lykins, A. D., & Cantor, J. M. (2017). Leg clinical assessment. Archives of Sexual Behavior, 32(6), 573–581. length versus torso length in pedophilia: Further evidence of https:// doi. org/ 10. 1023/A: 10260 93612 434 atypical physical development early in life. Sexual Abuse, 29(5), Bogaert, A. F. (2001). Handedness, criminality, and sexual offend- 500–514. https:// doi. org/ 10. 1177/ 10790 63215 609936 ing. Neuropsychologia, 39(5), 465–469. https:// doi. org/ 10. 1016/ Fazio, R. L., Lykins, A. D., & Cantor, J. M. (2014). Elevated rates S0028- 3932(00) 00134-2 of atypical handedness in paedophilia: Theory and implications. Boscarino, J. A., & Hoffman, S. N. (2007). Consistent association Laterality, 19(6), 690–704. https:// doi. org/ 10. 1080/ 13576 50X. between mixed lateral preference and PTSD: Confirmation among 2014. 898648 a national study of 2490 US Army Vietnam veterans. Psychoso- Fink, B., Neave, N., Laughton, K., & Manning, J. T. (2006). Second matic Medicine, 69(4), 365–369. https:// doi. or g/ 10. 1097/ PSY . to fourth digit ratio and sensation seeking. Personality and Indi- 0b013 e3180 5fe2bc vidual Differences, 41(7), 1253–1262. https:// doi. org/ 10. 1016/j. Brower, M. C., & Price, B. H. (2001). Neuropsychiatry of frontal lobe paid. 2006. 05. 002 dysfunction in violent and criminal behaviour: A critical review. Gerwinn, H., Weiß, S., Tenbergen, G., Amelung, T., Födisch, C., Pohl, Journal of Neurology, Neurosurgery & Psychiatry, 71(6), 720– A., Massau, C., Kneer, J., Mohnke, S., Kärgel, C., Wittfoth, M., 726. https:// doi. org/ 10. 1136/ jnnp. 71.6. 720 Jung, S., Drumkova, K., Schiltz, K., Walter, M., Beier, K. M., Cantor, J. M., Blanchard, R., Christensen, B. K., Dickey, R., Klassen, Walter, H., Ponseti, J., Schiffer, B., & Kruger, T. H. C. (2018). P. E., Beckstead, A. L., Blak, T., & Kuban, M. E. (2004). Intelli- Clinical characteristics associated with paedophilia and child sex gence, memory, and handedness in pedophilia. Neuropsychology, offending—Differentiating sexual preference from offence status. 18(1), 3–14. European Psychiatry, 51, 74–85. https://doi. or g/10. 1016/j. eur psy. Cantor, J. M., Blanchard, R., Robichaud, L. K., & Christensen, B. K. 2018. 02. 002 (2005a). Quantitative reanalysis of aggregate data on IQ in sexual Gorber, S. C., Tremblay, M., Moher, D., & Gorber, B. (2007). A compar- offenders. Psychological Bulletin, 131(4), 555–568. https:// doi. ison of direct vs. self-report measures for assessing height, weight org/ 10. 1037/ 0033- 2909. 131.4. 555 and body mass index: A systematic review. Obesity Reviews, 8(4), Cantor, J. M., Kabani, N., Christensen, B. K., Zipursky, R. B., Barbaree, 307–326. https:// doi. org/ 10. 1111/j. 1467- 789X. 2007. 00347.x H. E., Dickey, R., Klassen, P. E., Mikulis, D. J., Kuban, M. E., Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest peo- Blak, T., Richards, B. A., Hanratty, M. K., & Blanchard, R. (2008). ple in the world? Behavioral and Brain Sciences, 33(2–3), 61–83. Cerebral white matter deficiencies in pedophilic men. Journal of https:// doi. org/ 10. 1017/ S0140 525X0 99915 2X Psychiatric Research, 42(3), 167–183. https:// doi. org/ 10. 1016/j. Hubert, M., & Vandervieren, E. (2008). An adjusted boxplot for skewed jpsyc hires. 2007. 10. 013 distributions. Computational Statistics & Data Analysis, 52(12), Cantor, J. M., Klassen, P. E., Dickey, R., Christensen, B. K., Kuban, M. 5186–5201. https:// doi. org/ 10. 1016/j. csda. 2007. 11. 008 E., Blak, T., Williams, N. S., & Blanchard, R. (2005b). Handedness Ihme, J. M., Lemke, F., Lieder, K., Martin, F., Müller, J. C., & Schmidt, in pedophilia and hebephilia. Archives of Sexual Behavior, 34(4), S. (2009). Comparison of ability tests administered online and in 447–459. https:// doi. org/ 10. 1007/ s10508- 005- 4344-7 the laboratory. Behavior Research Methods, 41(4), 1183–1189. Cantor, J. M., Kuban, M. E., Blak, T., Klassen, P. E., Dickey, R., & https:// doi. org/ 10. 3758/ BRM. 41.4. 1183 Blanchard, R. (2007). Physical height in pedophilic and hebephilic Imhoff, R., Schmidt, A. F., Nordsiek, U., Luzar, C., Young, A. W., sexual offenders. Sexual Abuse, 19(4), 395–407. https://doi. or g/10. & Banse, R. (2010). Viewing Time effects revisited: Prolonged 1177/ 10790 63207 01900 405 response latencies for sexually attractive targets under restricted Condon, D. M., & Revelle, W. (2014). The international cognitive abil- task conditions. Archives of Sexual Behavior, 39(6), 1275–1288. ity resource: Development and initial validation of a public-domain https:// doi. org/ 10. 1007/ s10508- 009- 9595-2 measure. Intelligence, 43, 52–64. https:// doi. org/ 10. 1016/j. intell. Jahnke, S., & Malón, A. (2019). How pedohebephilic men think about 2014. 01. 004 adult-child sex: Effects of child gender and physical maturity. Psy- Condon, D. M., & Revelle, W. (2015). Selected ICAR data from the chology, Crime & Law, 25(1), 90–107. https:// doi. org/ 10. 1080/ SAPA-Project: Development and initial validation of a public-10683 16X. 2018. 15036 65 domain measure [Data set]. https:// doi. or g/ 10. 7910/ D VN/ Jahnke, S., Schmidt, A. F., Geradt, M., & Hoyer, J. (2015). Stigma- AD9RVY related stress and its correlates among men with pedophilic sexual Denny, K. (2009). Handedness and depression: Evidence from a large interests. Archives of Sexual Behavior, 44(8), 2173–2187. https:// population survey. Laterality, 14(3), 246–255. https:// doi. org/ 10. doi. org/ 10. 1007/ s10508- 015- 0503-7 1080/ 13576 50080 23628 69 Jahnke, S., Schmidt, A. F., & Hoyer, J. (2021). Perceived non-coercive Denny, K. (2011). Handedness and drinking behaviour. British Journal childhood sexual experiences among community men with and of Health Psychology, 16(2), 386–395. https:// doi. org/ 10. 1348/ without pedohebephilia in two non-matched case-control stud- 13591 0710X 515705 ies [Manuscript submitted for publication]. Department of Health Promotion and Development, University of Bergen. Dombert, B., Schmidt, A. F., Banse, R., Briken, P., Hoyer, J., Neutze, J., & Osterheider, M. (2016). How common is men’s self-reported 1 3 Archives of Sexual Behavior (2022) 51:849–866 865 Jordan, K., Wild, T. S. N., Fromberger, P., Müller, I., & Müller, J. L. abuse in a community sample of pedophiles and hebephiles. Sex- (2020). Are there any biomarkers for pedophilia and sexual child ual Abuse, 23(2), 212–242. https:// doi. org/ 10. 1177/ 10790 63210 abuse? A review. Frontiers in Psychiatry, 10, 940. https:// doi. org/ 382043 10. 3389/ fpsyt. 2019. 00940 Ogunfowokan, A. A., Olagunju, O. E., Olajubu, A. O., Faremi, F. A., Joyal, C. C., Black, D. N., & Dassylva, B. (2007). The neuropsychology Oloyede, A. S., & Sharps, P. W. (2016). Correlates of self-report of and neurology of sexual deviance: A review and pilot study. Sexual rape among male school adolescents in Ile-Ife, Nigeria. Journal of Abuse: A Journal of Research and Treatment, 19(2), 155–173. Interpersonal Violence, 31(4), 555–571. https:// doi. org/10 .11 77/ https:// doi. org/ 10. 1007/ s11194- 007- 9045-408862 60514 556104 Joyal, C. C., & Carpentier, J. (2021). Concordance and discordance Palta, M., Prineas, R. J., Berman, R., & Hannan, P. (1982). Comparison between paraphilic interests and behaviors: A follow-up study. of self-reported and measured height and weight. American Jour- Journal of Sex Research. https://doi. or g/10. 1080/ 00224 499. 2021. nal of Epidemiology, 115(2), 223–230. https:// doi. org/ 10. 1093/ 19868 01oxfor djour nals. aje. a1132 94 Joyal, C. C., Kärgel, C., Kneer, J., Amelung, T., Mohnke, S., Tenbergen, Pedneault, C. I., Hilgard, J., Pettersen, C., Hermann, C. A., White, K., G., Walter, H., & Kruger, T. H. C. (2019). The neurobiological & Nunes, K. L. (2021). How well do indirect measures assess origins of pedophilia: Not that simple. Journal of Sexual Medicine, sexual interest in children? A meta-analysis. Journal of Consult- 16(1), 153–154. https:// doi. org/ 10. 1016/j. jsxm. 2018. 10. 015 ing and Clinical Psychology, 89(4), 350–363. https:// doi. org/ 10. Lalumière, M. L., Blanchard, R., & Zucker, K. J. (2000). Sexual ori-1037/ ccp00 00627 entation and handedness in men and women: A meta-analysis. Poeppl, T. B., Eickhoff, S. B., Fox, P. T., Laird, A. R., Rupprecht, R., Psychological Bulletin, 126(4), 575–592. https://doi. or g/10. 1037/ Langguth, B., & Bzdok, D. (2015). Connectivity and functional 0033- 2909. 126.4. 575 profiling of abnormal brain structures in pedophilia. Human Brain Lehmann, R. J. B., Schmidt, A. F., & Jahnke, S. (2020). Stigmatiza- Mapping, 36(6), 2374–2386. https:// doi. org/ 10. 1002/ hbm. 22777 tion of paraphilias and psychological conditions linked to sexual Poeppl, T. B., Nitschke, J., Dombert, B., Santtila, P., Greenlee, M. W., offending. Journal of Sex Research, 58(4), 438–447. https:// doi. Osterheider, M., & Mokros, A. (2011). Functional cortical and org/ 10. 1080/ 00224 499. 2020. 17547 48 subcortical abnormalities in pedophilia: A combined study using Levenson, J. S., & Ackerman, A. R. (2017). The relationship between a choice reaction time task and fMRI. Journal of Sexual Medi- sex offender height and pedophilic interest. Deviant Behavior, 38, cine, 8(6), 1660–1674. https://doi. o rg/10. 1111/j. 1743- 610 9.2011. 1383–1392. https:// doi. org/ 10. 1080/ 01639 625. 2016. 12549 86 02248.x Lichtenberger, E. O., & Kaufman, A. S. (2009). Essentials of WAIS-IV Ray, J. J. (1984). The reliability of short social desirability scales. Jour- assessment. John Wiley & Sons. nal of Social Psychology, 123(1), 133. Liu, J. (2011). Early health risk factors for violence: Conceptualiza- Schiffer, B., Amelung, T., Pohl, A., Kaergel, C., Tenbergen, G., Ger - tion, evidence, and implications. Aggression and Violent Behavior, winn, H., Mohnke, S., Massau, C., Matthias, W., Weiß, S., Marr, 16(1), 63–73. https:// doi. org/ 10. 1016/j. avb. 2010. 12. 003 V., Beier, K. M., Walter, M., Ponseti, J., Krüger, T. H. C., Schiltz, Maechler, M., Rousseeuw, P., Croux, C., Todorov, V., Ruckstuhl, A., K., & Walter, H. (2017). Gray matter anomalies in pedophiles Salibian-Barrera, M., Verbeke, T., Koller, M., Conceicao, E. L. with and without a history of child sexual offending. Translational T., & di Palma, M. A. (2019). robustbase: Basic robust statistics Psychiatry, 7(5), e1129. https:// doi. org/ 10. 1038/ tp. 2017. 96 (0.93–5) [Computer software]. https://CRAN .R-pr oject. or g/pac ka Schiffer, B., Peschel, T., Paul, T., Gizewski, E., Forsting, M., Leygraf, ge= robus tbase N., Schedlowski, M., & Krueger, T. H. C. (2007). Structural brain Massau, C., Tenbergen, G., Kärgel, C., Weiß, S., Gerwinn, H., Pohl, abnormalities in the frontostriatal system and cerebellum in pedo- A., Amelung, T., Mohnke, S., Kneer, J., Wittfoth, M., Ristow, I., philia. Journal of Psychiatric Research, 41(9), 753–762. https:// Schiltz, K., Beier, K. M., Ponseti, J., Walter, M., Kruger, T. H. doi. org/ 10. 1016/j. jpsyc hires. 2006. 06. 003 C., Walter, H., & Schiffer, B. (2017). Executive functioning in Schipolowski, S., Wilhelm, O., Schroeders, U., Kovaleva, A., Kemper, pedophilia and child sexual offending. Journal of the International C. J., & Rammstedt, B. (2014). Eine kurze Skala zur Messung- Neuropsychological Society, 23(6), 460–470. https:// doi. org/ 10. kristalliner Intelligenz: Die Kurzskala gc des Berliner Tests zur 1017/ S1355 61771 70003 15 Erfassung Fluider und Kristalliner Intelligenz (BEFKI GC-K). McKinlay, A., Grace, R. C., McLellan, T., Roger, D., Clarbour, J., & GESIS-Working Papers 2014/29. http://nbn- r esolving. de/ ur n:nbn: MacFarlane, M. R. (2014). Predicting adult offending behavior for de: 0168- ssoar- 377864 individuals who experienced a traumatic brain injury during child- Schmidt, A. F., Babchishin, K. M., & Lehmann, R. J. B. (2017). A meta- hood. Journal of Head Trauma Rehabilitation, 29(6), 507–513. analysis of viewing time measures of sexual interest in children. https:// doi. org/ 10. 1097/ HTR. 00000 00000 000000 Archives of Sexual Behavior, 46(1), 287–300. https:// doi. org/ 10. McPhail, I. V., & Cantor, J. M. (2015). Pedophilia, height, and the 1007/ s10508- 016- 0806-3 magnitude of the association: A research note. Deviant Behavior, Schmidt, A. F., Mokros, A., & Banse, R. (2013). Is pedophilic sexual 36(4), 288–292. preference continuous? A taxometric analysis based on direct and Merz, Z. C., Lace, J. W., & Eisenstein, A. M. (2020). Examining broad indirect measures. Psychological Assessment, 25(4), 1146–1153. intellectual abilities obtained within an mTurk internet sample. https:// doi. org/ 10. 1037/ a0033 326 Current Psychology. https://doi. or g/10. 1007/ s12144- 020- 00741-0 Seto, M. C. (2008). Pedophilia and sexual offending against children: Miller, G. A., & Chapman, J. P. (2001). Misunderstanding analysis Theory, assessment, and intervention (1st ed.). American Psycho- of covariance. Journal of Abnormal Psychology, 110(1), 40–48. logical Association. https:// doi. org/ 10. 1037/ 11639- 000. https:// doi. org/ 10. 1037// 0021- 843x. 110.1. 40 Seto, M. C. (2004). Pedophilia and sexual offenses against children. Moffitt, T. E., Lynam, D. R., & Silva, P. A. (1994). Neuropsychological Annual Review of Sex Research, 15(1), 321–361. https:// doi. org/ tests predicting persistent male delinquency. Criminology, 32(2), 10. 1080/ 10532 528. 2004. 10559 823 277–300. https:// doi. org/ 10. 1111/j. 1745- 9125. 1994. tb011 55.x Seto, M. C. (2012). Is pedophilia a sexual orientation? Archives Mokros, A., Habermeyer, E., & Küchenhoff, H. (2018). The uncertainty of Sexual Behavior, 41(1), 231–236. https:// doi. or g/ 10. 1007/ of psychological and psychiatric diagnoses. Psychological Assess-s10508- 011- 9882-6 ment, 30(4), 556–560. https:// doi. org/ 10. 1037/ pas00 00524 Singy, P. (2015). Hebephilia: A postmortem dissection. Archives of Sexual Behavior, 44(5), 1109–1116. https:// doi. or g/ 10. 1007/ Neutze, J., Seto, M. C., Schaefer, G. A., Mundt, I. A., & Beier, K. M. s10508- 015- 0542-0 (2011). Predictors of child pornography offenses and child sexual 1 3 866 Archives of Sexual Behavior (2022) 51:849–866 Stephens, S., Seto, M. C., Goodwill, A. M., & Cantor, J. M. (2017). Welsch, R., Schmidt, A. F., Turner, D., & Rettenberger, M. (2021). Test– Evidence of construct validity in the assessment of hebephilia. retest reliability and temporal agreement of direct and indirect Archives of Sexual Behavior, 46(1), 301–309. https:// doi. org/ 10. sexual interest measures. Sexual Abuse, 33(3), 339–360. https:// 1007/ s10508- 016- 0907-zdoi. org/ 10. 1177/ 10790 63220 904354 Tanner, J. M. (1990). Foetus into man: Physical growth from conception Young, S. R., & Keith, T. Z. (2020). An examination of the convergent to maturity. Harvard University Press. validity of the ICAR16 and WAIS-IV. Journal of Psychoeduca- Tenbergen, G., Wittfoth, M., Frieling, H., Ponseti, J., Walter, M., Wal- tional Assessment, 38(8), 1052–1059. https:// doi. or g/ 10. 1177/ ter, H., Beier, K. M., Schiffer, B., & Kruger, T. H. C. (2015). The 07342 82920 943455 neurobiology and psychology of pedophilia: Recent advances and Yule, M. A., Brotto, L. A., & Gorzalka, B. B. (2014). Biological markers challenges. Frontiers in Human Neuroscience. https:// doi. org/ 10. of asexuality: Handedness, birth order, and finger length ratios in 3389/ fnhum. 2015. 00344 self-identified asexual men and women. Archives of Sexual Behav - Tenconi, E., Santonastaso, P., Degortes, D., Bosello, R., Titton, F., ior, 43(2), 299–310. https:// doi. org/ 10. 1007/ s10508- 013- 0175-0 Mapelli, D., & Favaro, A. (2010). Set-shifting abilities, central coherence, and handedness in anorexia nervosa patients, their unaf- Publisher's Note Springer Nature remains neutral with regard to fected siblings and healthy controls: Exploring putative endophe- jurisdictional claims in published maps and institutional affiliations. notypes. World Journal of Biological Psychiatry, 11(6), 813–823. https:// doi. org/ 10. 3109/ 15622 975. 2010. 483250 Pacific Psychological Assessment Cooperation. (2004). The Not-Real People stimulus set for assessment of sexual interest. Author. 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Archives of Sexual Behavior Springer Journals

Neurodevelopmental Differences, Pedohebephilia, and Sexual Offending: Findings from Two Online Surveys

Loading next page...
 
/lp/springer-journals/neurodevelopmental-differences-pedohebephilia-and-sexual-offending-HzG6SDS0ue

References (85)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2022
ISSN
0004-0002
eISSN
1573-2800
DOI
10.1007/s10508-021-02228-w
Publisher site
See Article on Publisher Site

Abstract

The neurodevelopmental theory of pedohebephilia states that sexual interests in children arise from early neurodevelopmental perturbations, as, for example, evidenced by increased non-right-handedness, more childhood head injuries, and reduced intelligence and height. As corroborating evidence largely rests on samples of convicted men, we conducted online surveys among German-speaking (Study 1, N = 199) and English-speaking men (Study 2, N = 632), specifically targeting community members with pedohebephilic or teleiophilic interests. Although we detected theoretically meaningful sexual interest pat- terns in an embedded viewing time task, we could not detect expected neurodevelopmental differences between teleiophilic and pedohebephilic men in either of the two studies. Strikingly, pedohebephilic men who reported convictions for sexual offenses emerged as shorter and less intelligent than pedohebephilic men without convictions in Study 2. While elucidating possible third variable confounds, results have to be interpreted cautiously because of the methodological problems inherent to non-matched case control designs. Keywords Pedophilia · Child sexual abuse · Neurobiology · Neurodevelopment · Etiology · DSM-5 Introduction Cantor et al., 2007, 2005a,b; Fazio, 2018). Yet, as the majority of studies are mostly restricted to samples of men who have In order to solve the puzzle of what may cause pedophilia or sexually offended, it is problematic that some of these factors hebephilia, that is, adult men's sexual attraction to sexually are also known markers of criminal behavior (e.g., lower height immature pre- or pubescent partners, respectively (Seto, 2012), and intelligence, Beckley et al., 2014; non-right-handedness, numerous studies have compared neurodevelopmental differ - Bogaert, 2001; early head injuries, Liu, 2011). Adding the fact ences between participants with and without pedophilia (see that there are many men with pedohebephilic interests who Tenbergen et al., 2015 for an overview). For example, neuroim- refrain from offending and are socially well adapted (Joyal aging techniques revealed structural or functional differences in et al., 2019), it is questionable whether these differences really the brains of pedophilic and teleiophilic (i.e., sexually attracted are due to patterns of sexual attraction and not to other fac- to adults) participants (e.g., Cantor et al., 2008; Poeppl et al., tors that are associated with (having been or) being arrested 2011, 2015). Furthermore, several studies found an association or convicted. To rule out such an alternative interpretation, a between pedophilia and soft markers for early neurodevelop- replication of neurodevelopmental differences between par - mental perturbations, such as non-right-handedness, height, IQ, ticipants with and without pedophilia in community samples or head injuries before age 13 (Blanchard et al., 2002, 2007; seems mandatory (Joyal et al., 2007). For solving the cardinal problem of recruiting non-insti- tutionalized participants with an uncommon and stigmatized * Sara Jahnke sexual attraction pattern such as pedophilia or hebephilia, the sara.jahnke@uib.no anonymity of online studies might be particularly helpful. Department of Health Promotion and Development, Following this logic, we sought to compare markers for neu- University of Bergen, 5020 Bergen, Norway rodevelopmental differences among men with different sexual Department of Psychology, Social & Legal Psychology, maturity interests in German-speaking (Study 1) and English- Johannes Gutenberg-Universität Mainz, Mainz, Germany speaking (Study 2) online samples. With a combined sample Institute for Clinical Psychology and Psychotherapy, of 831 participants, we hope to extend the currently limited Technische Universität Dresden, Dresden, Germany Vol.:(0123456789) 1 3 850 Archives of Sexual Behavior (2022) 51:849–866 evidence base concerning the relationship between neurodevel- to a pubescent and finally a postpubescent state (Blanchard opmental differences, pedohebephilia, and offending behavior. et al., 2007; see also Stephens et al., 2017 for a discussion of similar arousal patterns among hebephilic and pedophilic A Critical Discussion of the Early men who have sexually offended). Empirically, hebephilic Neurodevelopmental Perturbations Hypothesis men have sometimes been found to show IQ scores, rates of right-handedness, and childhood brain injuries that are Currently, pedohebephilic interest is increasingly viewed intermediate between those obtained among pedophilic and as being caused by an “underlying brain dysfunction, one teleiophilic men in forensic samples (Blanchard et al., 2003, that prevented the development of more typical intellectual 2007; Cantor et al., 2005a, 2005b). and sexual characteristics” (Cantor et al., 2005a, 2005b, p. 448). Neurodevelopmental deficits can be caused by genetic Potential Biases Associated with Studying disorders, accidents, brain tumors, and pre- or postnatal Pedohebephilia in Clinical or Forensic Samples exposure to teratogenic/toxic substances in utero (Becerra García, 2009). While a variety of such perturbations can Prior studies attempted to control for factors associated with have severe and long-lasting effects on cognitive function - criminality by comparing people with pedohebephilia with ing, they also manifest as "very mild, often inconsequential teleiophilic people, who have also mostly been referred features that are associated with atypical neurodevelopment" because of criminal or disturbing sexual activities (e.g., Blan- (Fazio, 2018, p. 1205), such as non-right-handedness, unde- chard et al., 2002; Fazio et al., 2017). Yet, just as increased tached ear lobes, or other biomarkers (see Jordan et al., 2020 levels of antisociality among teleiophilic men who have sexu- for a critical overview). Empirical research supporting the ally offended would not lead us to suppose that being anti- neurodevelopmental perturbations hypothesis of pedohebe- social is a characteristic feature of teleiophilia, the problem philia has mostly been conducted on men who have been remains that the detected markers for neurodevelopmental referred to clinical institutions "as a result of illegal or clini- perturbations may be more relevant for pedohebephilic men cally significant sexual behaviors or interests" (Cantor et al., who sexually offended than for pedohebephilia per se. Some 2007, p. 397). Based on those samples, findings from several researchers have therefore cautioned that "it seems far-fetched neuroimaging studies indicated structural abnormalities in to associate pedophilia with neurodevelopmental disorders" the brains of pedophilic men who have sexually offended (Joyal et al., 2019, p. 154) as long as unequivocal empirical compared to forensic or non-forensic teleiophilic controls support for such an account is lacking. (Cantor et al., 2008; Schiffer et al., 2007). Researchers have Additionally, some neurodevelopmental markers dis- also reported higher rates of non-right-handedness (Blan- cussed as being specific for pedophilia, such as increased chard et al., 2007; Cantor et al., 2005a, 2005b), lower height rates of left handedness/ambidexterity, have also been found (Cantor et al., 2007; Fazio et al., 2017), more head injuries among men with non-heterosexual orientation (Lalumière during childhood (Blanchard et al., 2002, 2003), more minor et al., 2000) or asexuality (Yule et al., 2014). Moreover, physical abnormalities (Dyshniku et al., 2015), and lower non-right-handedness has been linked to alcohol consump- IQs (Blanchard et al., 2007) among forensic/clinical samples tion (Denny, 2011) and a plethora of other psychological of pedohebephilic men. Notably, data on soft markers for disorders such as posttraumatic stress disorder (Boscarino neurodevelopmental perturbations especially need replica- & Hoffman, 2007 ), depression (Denny, 2009), or anorexia tion, because they were mostly obtained in the same insti- nervosa (Tenconi et al., 2010), to name a few. Lower IQ, tution (Centre for Addiction and Mental Health, Toronto) lower fraternal birth order, neuropsychiatric abnormalities, from members of the Kurt Freund Laboratory. All studies non-right-handedness, and lower D2/D4 length are also asso- were conducted on (potentially overlapping samples of) cases ciated with sexual offending in general, non-sexual delin- referred mostly for forensic evaluations (see, e.g., Blanchard quency, thrill seeking proclivity, and aggressive traits (Bailey et al., 2002), precluding the recruitment of larger numbers of & Hurd, 2005; Bogaert, 2001; Brower & Price, 2001; Fink pedohebephilic men who have not offended. et al., 2006; Moffitt et al., 1994; Ogunfowokan et al., 2016). Hebephilia is not a universally accepted concept, and the These findings underscore the unspecificity of the neurode- proposition to include sexual interests in pubescents as a dis- velopmental markers that have been associated with pedo- order category was not adopted in the DSM-5 (Singy, 2015). hebephilia, and invite speculation whether the differences Yet, irrespective of whether hebephilic interests can be con- between pedohebephilic men who have sexually offended and sidered pathological, the concept has demonstrated its merit teleiophilic control groups may be attributable to variables for research purposes. There are good reasons to assume that other than sexual interest. The same problems apply to stud- sexual attraction to prepubescents and to pubescents have ies that compare pedohebephilic men from treatment settings similar causes, since the course of sexual maturation leads to with healthy teleiophilic men, as the former group is likely gradual changes in physical appearance from a pre-pubescent to experience more distress about their sexuality or more 1 3 Archives of Sexual Behavior (2022) 51:849–866 851 concerns about sexual offending than pedohebephilic men The Present Studies who did not seek help. In a recent Swedish study, Abé et al. (2021) compared 55 help-seeking men with pedophilic dis- In the following studies, we sought to test selected indicators order from a national helpline with 57 age-matched controls of neurodevelopmental differences within two community recruited via the institute homepage and university services. samples, focusing solely on markers that are accessible in an The authors also report significant differences between the anonymous online setting. Prior studies on the neurodevelop- brains of pedophilic patients and controls, which, for the most mental correlates of pedophilia among forensic samples typi- part, remained significant when excluding patients who had cally classified participants as pedohebephilic either based on committed either sexual abuse or child pornography offenses, offense behavioral indicators as proxies (e.g., Cantor et al., indicating that these findings are not explained solely by 2005a, 2005b) or on participants’ responses in a phallometric criminality or antisociality. Furthermore, pedophilic disor- test or their self-reported sexual interests (e.g., Blanchard der was linked to a lower IQ, higher rates of several mood et al., 2002, 2003; Fazio et al., 2017). In the present studies, and anxiety disorders, and antisocial personality disorders, we employed self-report and viewing time (VT) measures to while the study found no differences regarding handedness assess sexual maturity interests. The VT measure was cho- and height (with pedophilic men emerging descriptively as sen as an objective measure (Schmidt et al., 2017) for cross- taller and less likely to be non-right-handed). validation of the self-report. By not relying on expert diagnosis of pedophilic disorder, Neurodevelopmental Perturbation Markers we avoided classifying participants based on categorical dis- in Community Samples of Pedohebephilic Men order diagnoses which in the case of pedophilia suffer from particularly high uncertainty (i.e., roughly between one- and About 50% or more of sexual offenses against children are two-thirds of pedophilia diagnoses may be considered wrong committed by teleiophilic men (Schmidt et al., 2013; Seto, due to low interrater reliability, Mokros et al., 2018). Our 2008), whereas many pedophilic men never commit sexual community data are also less likely to be confounded by third offenses at all (Dombert et al., 2016; Joyal & Carpentier, variables that are associated with caseness status. This is a 2021). Pedohebephilic men in community settings tend to typical problem of case–control designs, whereby individuals show better psychological adjustment than those recruited who by definition need to show clinically significant impair - in clinical or forensic environments (Jahnke et al., 2015). ment in terms of subjective distress or interpersonal function- Recently, an important first step toward disentangling fac- ing are compared to persons who are unlikely to exhibit signs tors associated with sexual offending and pedohebephilic of distress or functioning deficits. Yet, note that non-matched attraction has been made. Based on data from the German case–control studies such as the present ones are tainted by research project Neural Mechanisms underlying Pedophilia other potential biases, which we will discuss in limitations and Child Sexual Abuse (NeMUP), Gerwinn et al. (2018) section. found that "many of the factors reported as being related to paedophilia may […] actually be associated with commit- Hypotheses Above and beyond testing whether VT para- ting child sex offences and vice versa" (p. 75). In their sam- digms were valid in a specific online environment, we will ple, pedophilic and non-pedophilic men did not differ with test the following hypotheses based on the discussed limita- regard to handedness, accidents leading to unconsciousness tions of the empirical evidence for the neurodevelopmental in childhood, or IQ. Yet, with 155 pedophilic participants perturbation hypothesis: (about half of which had never committed sexual offenses), a larger evidence base is needed to achieve sufficient test (1) In case the neurodevelopmental perturbation hypoth- power to corroborate the purported small effect sizes of links esis will be specic fi for pedohebephilic sexual interests, between pedohebephilia and markers for neurodevelopmental pedohebephilic men will differ from teleiophilic men impairment, as the authors point out themselves. Regarding (i.e., men with a sexual interest in sexually mature indi- further indicators of neurodevelopmental differences, results viduals) with respect to markers of neurodevelopmental based on participants in the NeMUP project revealed simi- perturbations in community samples. lar results, namely that structural abnormalities in the brain (2) In case of a confound with self-reported sexual offend- (Schiffer et al., 2017) and impairments in executive function- ing status, pedohebephilic men who have sexually ing (Massau et al., 2017) are linked to offending status but offended will differ from pedohebephilic men who have not sexual interests (cf. Abé et al., 2021 who found that most not offended with respect to markers of neurodevelop- brain differences remained significant when controlling for mental perturbations. the sexual offending status of their pedophilic participants). We will test the following markers for neurodevelopmental perturbations: (a) height, (b) IQ, (c) non-right-handedness, 1 3 852 Archives of Sexual Behavior (2022) 51:849–866 Table 1 Sample size calculations Marker for neurodevelopmental deficits Source Effect size Comparison group Required sample size (two-tailed testing) Height McPhail and Cantor d = .20 Teleiophilic vs. pedohebephilic men 788 (2015) d = .21 Teleiophilic vs. pedophilic men (one-tailed: 620) (one-tailed: 564) Handedness Cantor et al. (2004) d = .25 Teleiophilic vs. pedohebephilic men 506 d = .50 Teleiophilic vs. pedophilic men 128 Head injuries before age 13 Blanchard et al. (2003) r = .12 None, pedophilic, hebephilic, and teleio- 542 philic participants were treated as an ordered set IQ (Wechsler Adult Intelligence Scale) Cantor et al. (2004) d = .32 Teleiophilic vs. pedohebephilic men 310 d = .53 Teleiophilic vs. pedophilic men 114 Calculations conducted with G*Power (for all calculations: 1–β = .80, α = .05, for height, handedness and IQ: t test, means: difference between two independent means, two groups: allocation ratio, N2/N1 = 1, for head injuries: exact, correlation, bivariate normal model: ρ H0 = 0, note that Spearman and Pearson correlations are computationally identical) Calculated from summary statistics in Table 1 and in-text (Cantor et al., 2004, p. 7), d = M –M / SD 1 2 pooled and (d) injuries resulting in unconsciousness before and after feasibility of the online design. Thus, we focused more on the age 13, all of which will be tested as separate hypotheses. descriptive point estimates of effect sizes than significance Study 1 was conducted among German-speaking individuals. levels. After we removed responses from 16 self-reported In Study 2, we improved the assessments of IQ and hand- female participants and two participants who did not report edness while repeating the study in a considerably larger their sex, the dataset contained 199 participants. Detailed English-speaking sample. information on how we categorized participants based on their sexual interests and offending status is presented in a separate section after the description of our measures. The Study 1 study was approved by the ethical review board of the Tech- nische Universität Dresden. Method Measures Participants and Procedure Participants filled out the scales in the order presented below. All participants were recruited on websites, blogs, and web We included additional questionnaires to assess interests forums to participate in a study on neurological develop- in sadomasochism, traumatic childhood experiences, and mental problems and traumatic childhood experiences. No non-negative sexual experiences in childhood that will be compensation was offered. Pedohebephilic individuals were featured in a separate publication (Jahnke et al., 2021). We sampled from websites for people with such sexual inter- were limited in our choice of intelligence measures, as most ests (jungsforum.net, schicksal-und-herausforderung.de, validated IQ scales are restricted to offline use by copyright krumme13.org, boylandonline.com, Deutsches Girllover- holders. Therefore, we used a scale to assess crystallized forum, ITP-Arcados), while teleiophilic participants were intelligence (Schipolowski et al., 2014) as the only validated recruited on science/psychology-related websites (e.g., German scale that was, to our knowledge, available for online forschung-erleben.uni-mannheim.de, caz-lesen.de, psychol- use at that time. We also developed a digit span test based ogie-heute.de, Facebook group "Psychologische Studien für on corresponding working memory tests in the Wechsler alle"). We attempted to collect as much information as pos- Adult Intelligence Scale (Lichtenberger & Kaufman, 2009). sible in a German-speaking sample. We stopped data col- Yet, because of severe validity concerns regarding these two lection when participation rates became very low. We did measures, we will not report their results. We calculated not attempt to reach a specific number of participants, as it The large majority (63%) of the participants scored between 10 and was clear to us prior to the study that a statistically sufficient 12 points on the crystallized intelligence test (note that the total score number of participants in German forums only will be unre- could theoretically range from 0 [no correct answer] to 12 [all items alistic (see Table 1 for sample size calculations). In contrast answered correctly]), indicating ceiling effects. Furthermore, a sizea- ble number of participants correctly recalled even 12-digit numbers on to the larger-scale Study 2, Study 1 was conducted to test the 1 3 Archives of Sexual Behavior (2022) 51:849–866 853 Cronbach’s α based on all of the 199 male participants in Head Injuries Participants indicated how many head injuries the dataset. leading to unconsciousness they had sustained before and after age 13. To prompt participant's memories, we trans- Viewing Time Measure of Sexual Interest We presented five lated and adapted questions from Blanchard et al. (2003), male and five female pictures of prepubescent children (Tan- replacing "hockey" with "soccer" as a more popular sports ner stage 1; Tanner, 1990) and postpubescent adults each activity among German children: "Were you ever knocked (Tanner stage 5) from the Not Real People Set (i.e., computer- unconscious during your childhood (before you were 13 years generated pictures of Caucasian individuals in bathing suits old), for example, falling from a tree or hitting your head in provided by the Pacific Psychological Assessment Coopera- a soccer game?" and "Were you ever knocked unconscious tion, 2004) in randomized order. Participants were prompted in adulthood (after 13 years of age), for example, in a car to indicate their attraction in a forced choice format with two accident or in a sports injury?". response alternatives within a one second response window: "Yes, this is a potential sexual partner for me" vs. "No, this is Social Desirability Tendencies to give socially desirable not a potential sexual partner for me" (i.e., speeded VT task, responses were assessed with an eight-item German scale by Imhoff et al., 2010, Study 3, but note that we did not exclude Ray (1984). Participants were asked to respond to questions trials with longer reaction times). We chose to instruct like “Are you quick to admit making a mistake?” with either speeded responding to prevent participants from producing "true" or "false." Higher values on the scale average indicate too many outliers due to unrestricted response windows in a higher risk for socially desirable responding (α = 0.67). an online survey. Response latencies were recorded unob- trusively. For further details on how we dealt with statistical Sociodemographic Information (Including Height and Previ‑ outliers see the section on VT classifications. Viewing time ous Convictions) We assessed participant sex (male, female, measures of pedohebephilia have repeatedly been shown to other), age, educational achievement, and height on a cm produce reliable (Welsch et al., 2021) and valid results (Ped- scale. One participant (with stronger self-reported sexual neault et al., 2021; Schmidt et al., 2017). attraction to adults than to children or adolescents) reported a height of 78 cm, which we corrected to the more likely Self‑Reported Sexual Interests A six-item scale introduced by value of 178 cm. Previous convictions for child sexual abuse, Jahnke and Malón (2019) was administered to assess pedo- rape, or possession of child sexual exploitation material were philic, hebephilic, and teleiophilic interests by collecting attrac- assessed on a binary scale (yes/no) with three separate items tion ratings on a Likert scale ranging from 1 (no sexual interest) (“I have been convicted for child sexual abuse,” “I have been to 10 (maximum sexual interest) with regard to children of each convicted for rape,” “I have been convicted for child pornog- sex before puberty, adolescents of each sex in early stages of raphy offenses”). puberty, and sexually mature adult men and women. Categorizing Participants’ Sexual Interests Adapted Edinburgh Handedness Inventory We used the and Oending Sta ff tus modified version of the Edinburgh Handedness Inventory (EHI, Dyshniku et al., 2015). Participants had to decide Classification Based on Self‑Reported Sexual Attraction whether they preferred the left or right hand for presented activities such as writing or drawing (forced choice). Due to In the pedophilic subsample, we retained only men who a technical mistake, it was not possible to choose both the left reported more sexual attraction to prepubescent children than and the right hand in cases where participants did not have to early-to-mid-pubertal adolescents or adults (by subtracting a clear preference (ambidexterity, note that this mistake was the maximum reported sexual attraction to either prepubertal fixed in Study 2 ). Like Dyshniku et al. (2015), we removed girls or boys from the maximum self-reported sexual attrac- the items “using a broom” and “opening a box” because of tion to either of the four other categories [pubescent girls and poor factor analytic loadings. Based on the responses to the boys, sexually mature men and women]). The teleiophilic eight remaining items (α = 0.96), we computed a laterality subsample consisted solely of men indicating highest sexual quotient by dividing the difference score of left and right attraction to adults (subtracting the maximum sexual attrac- responses through the sum of left and right responses. Higher tion to either sexually mature men or women from the maxi- scores indicate a stronger preference for the right hand. mum sexual attraction to either of the four remaining stimuli classes). Those whose sexual attraction was highest for early- to-mid-pubertal adolescents or equally high for pubescent and prepubescent stimuli were categorized as the hebephilic Footnote 1 (continued) subsample. Hence, the hebephilic group included seven indi- the working memory test, which raised doubts whether the digit span viduals with an equally strong attraction to prepubescent and trials were performed according to instructions. 1 3 854 Archives of Sexual Behavior (2022) 51:849–866 pubescent children. For the present analyses, the pedophilic and hebephilic group were combined into a pedohebephilic subsample. As categorization was based entirely on compari- sons of self-reported sexual interests to different groups of people, a difference of at least one unit between the stimulus categories on the sexual interest scale was sufficient for cate- gorization. Yet, one person in the teleiophilic group indicated a sexual interest in prepubescent children that surpassed the midpoint of the 10-point scale. Data from nine men who reported to have equally strong teleiophilic attraction and sexual attraction to one of the remaining four categories were excluded from the analyses. We collected data from 89 self-reported teleiophilic men and 101 self-reported pedohebephilic men. Table 2 gives an overview of self-reported maximum sexual attraction ratings as a function of sexual interest groups and sexual maturity status. The pedohebephilic group was then split up into two subgroups according to their self-reported status regarding previous convictions for sexual offending (child pornography offending and/or child sexual abuse, see Table  2 for more infor- mation on the nature of the convictions). In the teleiophilic group, only one person self-reported a sexual offense involv - ing children and was discarded from the analyses. The groups will be referred to as Pedo-SO (pedohebephilic men who do not report sexual offending), Pedo + SO (pedohebephilic men who do report sexual offending), and Tel-SO (teleiophilic men who do not report sexual offending) in results section. Hence, analyses based on sexual attraction and sexual offending status included 71 Pedo-SO, 30 Pedo + SO, and 88 Tel-SO. Classification Based on Viewing Time For 42 participants, we could not record any response laten- cies likely due to their use of anonymity software which did not permit the task to run properly. Response latency outliers were screened for each single trial utilizing the adjbox func- tion from the R package robustbase (Maechler et al., 2019). This procedure is based on a robust nonparametric modifi- cation of the standard Tukey criterion that specifically fits skewed distributions (Hubert & Vandervieren, 2008) which are notorious for response latency data. Empirically deter- mined outliers were set to missing values (6% of the trials). We calculated average response latencies for each Tanner by stimulus sex category, if we had recorded at least one value for each of the relevant stimulus categories (no cases were deleted because of outlier exclusion). We categorized groups based on the difference score between maximum response latencies to either male or female Tanner stage 5 and Tanner stage 1 stimuli. Similar difference coefficients have been shown to produce the most valid differentiation between teleiophilic and pedohebephilic groups (Schmidt et al., 2017). Participants with positive difference scores were classified as pedohebe- philic subsample (although technically reflecting pedophilic 1 3 Table 2 Self-reported maximum sexual attraction to female and male members of each category of sexual maturity, viewing time scores, and sexual offending status among the pedophilic, hebephilic, and teleiophilic group (Study 1, Study 2) Study 1 Study 2 a a Variable Pedophilia Hebephilia Teleiophilia Pedophilia Hebephilia Teleiophilia M (SD) n M (SD) n M (SD) M (SD) M (SD) n M (SD) n M (SD) n Self-reported attraction to prepubescents 9.57 (0.64) 53 6.00 (2.83) 48 1.35 (0.97) 89 9.74 (0.87) 137 6.77 (2.43) 141 1.30 (1.09) 317 Self-reported attraction to early–mid-pubescents 6.28 (2.05) 53 9.50 (0.83) 48 2.60 (1.95) 89 6.31 (2.31) 137 9.62 (0.76) 141 1.78 (1.71) 317 Self-reported attraction to adults 3.42 (2.48) 53 4.38 (2.45) 48 9.54 (1.02) 80 3.69 (2.46) 137 5.18 (2.54) 141 9.62 (0.91) 317 Viewing time score T1—T5 916 (970) 31 395 (769) 35 -685 (816) 84 - - - - - - Viewing time score T1,2,3- T4,5 - - - - - - 355 (869) 91 162 (781) 90 -831 (902) 313 Prior convictions for sexual offenses combined (in %) 26.4 53 33.3 48 1.1 89 14.6 137 13.5 141 1.3 317 Prior convictions for child sexual abuse (in %) 7.5 53 16.7 48 0.0 89 6.6 137 7.1 141 0.9 317 Prior convictions for rape (in %) 0.0 53 0.0 48 0.0 89 2.2 137 0.7 141 0.0 317 Prior convictions for child pornography offenses (in %) 20.8 53 25.0 48 1.1 89 12.4 137 7.8 141 0.3 317 Includes seven (Study 1) and 31 (Study 2) participants with an equally strong sexual attraction to prepubescents and early-to-mid-pubescents Archives of Sexual Behavior (2022) 51:849–866 855 Table 3 Planned comparisons (Helmert contrasts): Self-reported pedohebephilic men vs. teleiophilic men (Study 1) Variable Pedohebephilia (P –SO, Teleiophilia, no sexual (P-SO, P + SO) vs. T-SO P + SO) offending (T-SO) M (SD) n M (SD) n t (df) p d *** b Attraction to prepubescents 7.87 (2.68) 101 1.32 (0.93) 88 − 22.15 (75.86) < .001 -3.20 *** Attraction to early–mid-pubescents 7.81 (2.26) 101 2.53 (1.87) 88 − 17.00 (186) < .001 − 2.54 *** b Attraction to adults 3.87 (2.50) 101 9.55 (1.03) 88 18.76 (65.62) < .001 2.92 *** b Age 37.52 (12.44) 101 32.53 (11.45) 88 − 4.00 (104.31) < .001 − 0.42 *** Viewing time score T1—T5 640 (901) 66 -692 (818) 83 − 9.46 (146) < .001 − 1.57 Height 180.54 (6.55) 101 180.55 (6.52) 88 0.21 (186) .836 0.00 EHI Laterality Index 0.83 (0.48) 101 0.88 (0.43) 88 0.63 (186) .535 0.10 Head injuries before age 13 0.24 (0.67) 101 0.27 (0.64) 88 0.46 (186) .619 0.05 Head injuries after age 13 0.20 (0.69) 101 0.31 (0.81) 87 0.69 (185) .544 0.14 Social desirability 1.97 (0.56) 100 1.86 (0.50) 88 -1.47 (185) .144 − 0.20 *** p < .001 (two-sided) d = M –M / SD , calculated using the cohen.d function of the R package psych 1 2 pooled We used Welch’s correction due to unequal variances (as indicated by Levene test for equality of variances, center = median) p-value based on 1,000 bootstrap samples due to severe deviations from the assumption that residuals are normally distributed preferences for prepubescent children), while participants with hebephilic men showed positive average scores (indicating negative difference scores were coded as teleiophilic subsam- stronger sexual attraction to children at Tanner stage 1 than ple. Based on their VT profiles, we were able to categorize 67 adults at Tanner stage 5), while teleiophilic men achieved participants as pedohebephilic and 90 participants as teleio- negative scores (indicating stronger attraction to adults than philic. When considering offending status, this left us with 46 to children, see Table 2). VT-based Pedo-SO, 21 Pedo + SO, and 86 Tel-SO. Sample Description Statistical Procedure for Main Analyses Participants were rather well-educated (with 72%, 47%, and Reverse Helmert contrasts were employed for planned 82% reporting to have achieved University entrance cer- comparisons to assess whether there were any differences tificates in Germany for Pedo-SO, Pedo + SO, and Tel-SO, between Pedo-SO, Pedo + SO, and Tel-SO. Specifically, we respectively). Pedo + SO were significantly less educated compared Tel-SO with the mean of Pedo-SO/ + SO and Pedo- than Tel-SO (χ = 13.99, df = 1, p = 0.001; ϕ = 0.34), while SO with Pedo + SO. Because we had planned comparisons, we detected no difference between Pedo-SO and Tel-SO we bypassed omnibus analysis of variance. For some analy- (χ = 2.24, df = 1, p = 0.135; ϕ = 0.12). Pedo-SO/ + SO were ses, we opted against the use of parametric tests because the older than Tel-SO, and Pedo + SO were older than Pedo-SO distribution of residuals showed severe deviations from the (both significant, Tables  3, 4). Sample descriptions based on normal distribution or because of unequal variances (hetero- VT-inferred classifications can be found in Supplement A. scedasticity). In these cases, we used appropriate robust tests (bootstrapping or Welch corrections). Group Differences Unless stated otherwise, results reported here refer to classifi- Results cation based on self-report. Readers can access results based on VT-based classification in the Supplemental Materials. Agreement Between Self‑Reported and Viewing We could not detect significant differences for any of the clas- Time‑Inferred Sexual Interests sification procedures (see Tables  3, 4 for self-report-based classic fi ation and Table S1-S2 in Supplement B for VT-based The two classification procedures led to similar results for classification,), which was expected given the low statistical 127/150 (85%) men who were sorted concordantly as either power. Therefore, we put a stronger focus on effect size point teleiophilic or pedohebephilic based on self-report and VT estimates than the significance of the effects. (χ = 70.98, df = 1, φ = 0.69, p < 0.001). Pedophilic and 1 3 856 Archives of Sexual Behavior (2022) 51:849–866 Table 4 Planned comparisons (Helmert contrasts): Self-reported pedohebephilic men with vs. without convictions for sexual offending (Study 1) Variable Pedohebephilia, no sexual Pedohebephilia, sexual P-SO vs. P + SO offending (P –SO) offending (P + SO) M (SD) n M (SD) n t (df) p d Attraction to prepubescents 7.83 (2.77) 71 7.97 (2.48) 30 0.24 (60.60) .809 0.05 Attraction to early–mid-pubescents 7.63 (2.26) 71 8.23 (2.25) 30 1.32 (186) .188 0.27 Attraction to adults 3.92 (2.44) 71 3.77 (2.67) 30 − 0.26 (50.34) .794 − 0.06 *** b Age 34.03 (10.09) 71 45.80 (13.69) 30 4.25 (42.9) < .001 1.06 Viewing time score T1—T5 552 (879) 43 805 (938) 23 1.15 (146) .253 0.29 Height 180.85 (6.54) 71 179.83 (6.62) 30 − 0.71 (186) .479 − 0.16 EHI Laterality Index 0.83 (0.49) 71 0.83 (0.46) 30 − 0.01 (186) .994 0.00 Head injuries before age 13 0.25 (0.73) 71 0.20 (0.48) 30 − 0.38 (186) .652 − 0.08 Head injuries after age 13 0.16 (0.58) 71 0.30 (0.92) 30 0.84 (185) .463 0.20 Social desirability 1.95 (0.55) 71 2.01 (0.60) 29 0.52 (185) .601 0.11 *** p < .001 (two-sided) d = M –M / SD , calculated using the cohen.d function of the R package psych 1 2 pooled We used Welch’s correction due to unequal variances (as indicated by Levene test for equality of variances, center = median) p-value based on 1,000 bootstrap samples due to severe deviations from the assumption that residuals are normally distributed Further data inspection revealed that the Pedo + SO group head injuries, we determined significance based on 1,000 was about 1 cm shorter than the Pedo-SO and Tel-SO (see bootstrap samples because these variables showed severe Table 3). Our results also suggested a higher likelihood for deviations from the assumption of normality. There was no Pedo-SO/ + SO to be left-handed compared to Tel-SO (see significant relationship between age and head injuries before Table 3), while there was no indication for differences in age 13 (Pedo-SO: r = .07, 95%CI = [−.11, .40], Pedo + SO: handedness depending on sexual offending status among r = .20, 95%CI = [−.25, .50], Tel-SO: r = .01, 95%CI = [−.19, the pedohebephilic group (see Table 4). Furthermore, our .22]). We found, however, a significant negative relationship data indicated lower rates of head injuries before age 13 for between age and head injuries after age 13 for non-offending Pedo-SO/ + SO compared to Tel-SO (see Table 3). Again on a teleiophilic participants (Pedo-SO: r = .18, 95%CI = [−.18, descriptive level, Pedo + SO reported more head injuries after .47] Pedo + SO: r = .24, 95%CI = [−.16, 0.47], Tel-SO: age 13 than Pedo-SO (Table 4). All three groups achieved r = −.14, 95%CI = [−.25, -.003]). Similar non-effects were similar scores on the social desirability scale (Tables 3, 4). obtained for the VT-inferred classification, with the excep- In group comparisons based on VT-inferred classifications of tion of a significant negative link between height and age sexual orientation, we achieved results in similar directions, among VT-based Pedo + SO (see Supplement C). Taking with the sole difference that Pedo + SO were about 0.40 cm into account the number of tests, size, and direction of these taller than Pedo-SO, and that Pedo-SO/ + SO appeared more effects, these observations speak against a systematic pattern likely to be right-handed than Tel-SO (all ns., see Tables for the head injuries measure. Nevertheless, they indicate, S1—S2 in Supplement B). albeit not consistently, that differences in age may explain some of the (nonsignificant) differences for the contrasts Control Analyses based on height. As group differences in age might have biased our results, we assessed links between age, height, and head injuries before Discussion and after age 13 separately for each of the six tested subgroups (i.e., [1] self-reported Pedo-SO, Pedo + SO, and Tel-SO; [2] Study 1 showed that it is feasible—at least for a majority of VT-inferred Pedo-SO, Pedo + SO, and Tel-SO). For the self- the sampled individuals—to run an online version of the VT report classification, we found no significant link between measure among self-identified males with sexual interest in age and height (Pedo-SO: r = .08, p = .482, 95%CI = [−.15, children and adolescents. For the first time, these results extend .31], Pedo + SO: r = −.33, p = .072, 95%CI = [−.62, .03], Tel- the validation of VT tasks as measures of pedohebephilic sexual SO: r = −.13, p = .217, 95%CI = [−.33, .08]). Note that for interest (Schmidt et al., 2017) to self-identified pedohebephilic 1 3 Archives of Sexual Behavior (2022) 51:849–866 857 men from the community which can be regarded as the most countries of residences were allowed for participation (task direct test of VT validity hitherto. Nevertheless, results of Study language was English): Australia, Austria, Belgium, Can- 1 are subject to typical caveats associated with research based ada, Czech Republic, Denmark, Estonia, Finland, France, on self-reported data, retrospective questionnaires, and online Germany, Greece, Iceland, Italy, Lithuania, Luxembourg, research (see general discussion for a detailed elaboration). In Netherlands, New Zealand, Norway, Poland, Portugal, Slo- addition, statistical power was insufficient. Yet, descriptive vakia, Spain, Sweden, UK, and the USA. In sum, we recruited analyses indicated that differences between pedohebephilic and 331 participants via B4U-ACT and 318 via MTurk. After non-pedohebephilic groups might be driven by a confound of the exclusion of 17 people who did not disclose their sex or (self-reported) sexual offending status. Furthermore, pedohe- reported to be female or other, this left us with a total of 632 bephilic and teleiophilic men reported similar rates of head participants. Detailed information on how we categorized injuries before age 13, irrespective of their offending status. participants based on their sexual interests and offending Descriptively, our data appear to be more in line with research status is presented in a separate section after the description pointing toward a link between markers for neurodevelopmen- of our measures. tal differences and norm-breaking/criminal behavior, instead of pedohebephilia per se (with the exception of left-handedness, Measures which was more prominent in the pedohebephilic group) calling for a more stringent test based on adequate statistical power. Measurements were administered in the same order as they This was sought to achieve in Study 2. appear in the following section. Additionally, we assessed information about traumatic childhood experiences and non- negative sexual experiences in childhood for a separate pub- Study 2 lication (Jahnke et al., 2021). Method Viewing Time Measure of Sexual Interest We employed a similar VT task as in Study 1, again using stimulus material Participants and Sample Size Calculations from the Not Real People set (Pacific Psychological Assess- ment Cooperation, 2004). Deviating from Study 1, we added To reach more potential participants, which is necessary to pictures of early (Tanner stages 2–3) and late adolescents address the statistical power issue, and to make sure that sam- (Tanner stage 4), alongside pictures of adults (Tanner stage ples in Study 1 and 2 are (at least largely) independent, Study 5) and prepubescent children (Tanner stage 1). To shorten 2 targeted English-speaking men. We conducted sample size test duration, we presented only four pictures per sex and calculations for all markers for neurodevelopmental differ - Tanner stage category. ences based on effect sizes from the literature (see Table  1). We estimated that among all cognitive markers for neurode- Self‑Reported Sexual Interests We used the English version velopmental differences, self-reported height would show of the measure from Study 1. the smallest average difference between pedohebephilic and teleiophilic participants. Hence, we set our sampling goal to International Cognitive Ability Resource Sample Test To 620 participants, based on estimations with G*Power (setting measure cognitive ability, we used the 16-item sample test α = 0.05 and β = 0.20, one-sided testing for height, two-sided from the public-domain International Cognitive Ability testing for all other hypotheses, Faul et al., 2007). The study Resource (ICAR; Condon & Revelle, 2014), which consists of was approved by the institutional review board of the MSH letter and number series, matrix and verbal reasoning items, Medical School. and three-dimensional rotation items. Previous research has The study was advertised as a survey on wanted and established that the ICAR 16-item sample test is a reliable unwanted childhood sexual experiences, cognitive develop- and valid measure of global cognitive ability (Condon & ment, and sexual interests in children or adults among men Revelle, 2015; Merz et al., 2020). The (self-administered) from the community. We collected data via the B4U-ACT ICAR-16 shows high convergent validity with the Wechsler support group for people with pedohebephilia between July Adult Intelligence Scale (fourth edition) administered by 2018 and March 2019, offering to donate 1.50$ for each a trained clinician (r = 0.81 and r = 0.94 with the manifest participant to B4U-ACT (with a maximum donation sum and the latent score of the WAIS-IV, respectively, Young & of $300). We furthermore recruited participants from the Keith, 2020). As pedohebephilic participants might have a MTurk workforce. Only male MTurk workers who had been stronger motivation than teleiophilic MTurkers to excel on approved for 100 to 5000 human intelligence tasks with an the IQ test, it was planned to compare their responses with overall approval rate of 80% or more were eligible for the scores obtained among Internet users who wanted to test their study and received a payment of two US$. The following cognitive abilities in a previous study (Condon & Revelle, 1 3 858 Archives of Sexual Behavior (2022) 51:849–866 2015). To match procedures employed in this comparison men reported sexual interests surpassing the midpoint of the study, we presented the test without time restraints. For 18 respective scales. Group means for sexual attraction ratings participants (all recruited via B4U-ACT) we could not cal- are displayed in Table 2. In the teleiophilic group, four par- culate sum scales due to missing values. Internal consistency ticipants self-reported a sexual offense and were therefore was good (Kruder-Richardson 20 = 0.84). discarded from the T-SO group. Analyses based on sexual attraction and sexual offending status included 239 Pedo-SO, Adapted Edinburgh Handedness Inventory We used the 39 Pedo + SO, and 313 Tel-SO. English version of the EHI (Dyshniku et al., 2015) employed in Study 1, again without the broom and box items (α = 0.89, Classification Based on Viewing Time For 110 participants, no missing values). we could not record any response latency data, most likely due to the use of anonymity software. Using the same method Head Injuries As in Study 1, we used items from Blanchard to identify outliers as in Study 1, 4% of recorded trials were et al. (2013) to assess head injuries involving unconscious- marked as outliers. All participants had to have at least one ness before and after age 13. One participant recruited via valid response for each Tanner stage by stimulus sex category B4U-ACT left these questions unanswered. (note that this requirement was fulfilled in all cases where VT scores were recorded). Viewing time data were analyzed in Sociodemographic Information (Including Height and Previ‑ the same way as described in Study 1, with one exception: ous Convictions) As in Study 1, we assessed sex, age, and As Study 2 included pictures from Tanner stages 1 to 5, we educational achievement (i.e., having or not having obtained additionally calculated difference scores between maximum a Bachelor's degree or higher). Participants were allowed to average response latencies to either male and female stimuli choose whether to report their height in the metric (cm) or separately for each of the five Tanner stages. To determine the imperial system (the latter were subsequently transferred pedohebephilia, we subtracted maximum average reaction to cm). For height, we corrected obvious typing mistakes times to either male or female stimuli in Tanner stages 4 and 5 in four cases (e.g., 88 inches were changed to 8 inches). To (late pubescence and adulthood) from Tanner stages 1, 2, and determine outliers, we estimated fences based on the Tukey 3 (pre-peri-pubescence). Participants with positive difference criterion (1.5 interquartile ranges). This led to the exclusion scores were classified as pedohebephilic, while participants of five data points above 198.12 cm and six below 157.48 cm. with negative difference scores were classified as teleiophilic. Cases with heights outside the fences were about equally Thus, we categorized 60 participants as pedophilic (i.e., hav- split between participants from MTurk (2 above, 3 below) ing higher scores for Tanner stage 1 than for Tanner stages 2 and B4U-ACT (3 above, 3 below). Twelve participants from to 5), 119 as hebephilic (i.e., having higher values for Tanner B4U-ACT and two from MTurk did not report their height. stages 2 and 3 than for Tanner stages 1, 4, and 5), and 326 as For the assessment of previous convictions, we administered teleiophilic (i.e., having higher values for Tanner stages 4 and the same items as in Study 1. 5 than for Tanner stages 1 to 3). When accounting for offend- ing status previous convictions for sexual offending (child Categorizing Participants’ Sexual Interests and Oending ff pornography offending, rape, and/or child sexual abuse, see Status Table 2 for further information), 154 were categorized as Pedo-SO, 25 as Pedo + SO, and 326 as Tel-SO. Classification Based on Self‑Reported Sexual Attraction We formed groups based on self-reported sexual interests fol- lowing procedures in Study 1. This led us to categorize 317 participants as teleiophilic men and 278 as pedohebephilic Results men (141 with a predominant pedophilic interest). Twenty- one participants from the self-reported teleiophilic group Agreement Between Self‑Reported and Viewing were recruited via B4U-ACT, while seven participants Time‑Inferred Sexual Interests from the self-reported pedohebephilic group were sampled on MTurk. Those participants were not excluded from the For the 493 participants that we could classify based on self- analyses. Thirty-seven participants who reported equal reported and VT-inferred sexual interests, classification pro- maximum sexual attraction to adult and prepubertal/early- cedures led to similar results (χ = 123.32, df = 1, φ = 0.50, to-mid-pubertal persons were excluded. Among self-reported p < 0.001). Of the self-reported teleiophilic participants, 267 teleiophilic participants, only seven (2%) reported a sexual (85%) showed a teleiophilic pattern in the VT task, while 114 interest in prepubescent girls or boys surpassing the mid- (63%) of self-reported pedohebephilic participants showed point of the sexual attraction scale. Regarding girls or boys in VT results indicating pedohebephilic interests. The pedo- early-to-mid-puberty, only 20 (6%) self-reported teleiophilic philic and hebephilic groups achieved on average positive 1 3 Archives of Sexual Behavior (2022) 51:849–866 859 Table 5 Planned comparisons (Helmert contrasts): Self-reported pedohebephilic men vs. teleiophilic men (Study 2) Variable Pedohebephilia (P –SO, Teleiophilia, no sexual (P-SO, P + SO) vs. T-SO P + SO) offending (T-SO) M (SD) n M (SD) n t (df) p d *** b Attraction to prepubescents 8.23 (2.36) 278 1.29 (1.09) 313 − 38.97 (75.36) < .001 − 3.87 *** b Attraction to early–mid-pubescents 7.99 (2.38) 278 1.77 (1.71) 313 − 30.31 (92.57) < .001 − 3.04 *** b Attraction to adults 4.44 (2.61) 278 9.64 (0.91) 313 23.15 (53.09) < .001 2.73 Age 34.44 (12.98) 273 35.15 (11.06) 313 − 1.21 (113.35) .228 0.06 *** Viewing time score T1,2,3—T4,5 259 (830) 181 − 839 (904) 309 − 10.77 (487) < .001 − 1.25 Height 178.89 (7.55) 259 177.66 (7.18) 307 0.11 (563) .545 − 0.17 *** d EHI Laterality Index 0.68 (0.61) 278 0.68 (0.58) 313 − 0.77 (588) .387 0.00 ** b ICAR 9.86 (4.18) 260 7.24 (3.52) 313 − 3.23 (78.43) .002 − 0.69 Head injuries before age 13 0.25 (0.66) 277 0.25 (0.68) 313 − 0.20 (587) .867 0.00 b d Head injuries after age 13 0.20 (0.64) 276 0.30 (0.78) 313 0.01 (64.35) .989 0.14 *** ** p < .001, p < .01 (two-sided) d = M –M / SD , calculated using the cohen.d function of the R package psych 1 2 pooled We used Welch’s correction due to unequal variances (as indicated by Levene test for equality of variances, center = median) One-sided p p-value based on 1,000 bootstrap samples due to severe deviations from the assumption that residuals are normally distributed VT mean scores (indicating a stronger attraction to pedohe- intelligent than P + SO (Table 6). Furthermore, unobtrusively bephilic stimuli on Tanner stage 1–3 than to late/postpubes- recorded response latencies during the IQ test revealed that cent stimuli on Tanner stage 4 or 5), while the self-reported the Pedo-SO group had an almost twice as long average teleiophilic group showed negative VT scores (indicating a completion time as the Pedo + SO and the Tel-SO group stronger relative attraction to physically mature adults, see (with median scores of 15.09 min compared to 7.65 min and Table 2). 8.40 min, based on reaction time data from 153 Pedo-SO, 29 Pedo + SO, and 310 Tel-SO). To compare results with a sample that was arguably more motivated to score high, we Sample Description used data from the ICAR project (Condon & Revelle, 2015) to extract means and standard deviations for each of the items Educational levels were relatively high as 53%, 49%, and from the ICAR 16-item sample test among men over 18 years 62% indicated to have achieved an associate degree, BA (n between 4220 and 13,216, average n = 9324, note that the degree, or higher among Pedo-SO, Pedo + SO, and Tel-SO, ICAR items were delivered untimed in both our survey and respectively. We detected no significant difference between the ICAR project). All of these participants had solved a var- Tel-SO, Pedo-SO, and Pedo + SO regarding educational level ying number of the ICAR item set online in order to receive (χ = 5.27, df = 2, φ = 0.09, p = 0.072). Sexual offenses were a personalized feedback (Condon & Revelle, 2014). These reported mostly by pedohebephilic participants (Table 2). individual means and standard deviations were added to cre- Age did not differ significantly between Pedo-SO/ + SO and ate average sum scores (M = 8.60, SD = 7.42). t-tests with Tel-SO, but Pedo + SO were significantly older than Pedo-SO Welch-adjusted degrees of freedom based on these summary (Tables 5–6). Sample descriptions based on VT are presented statistics revealed that Pedo-SO/ + SO (M = 9.86, SD = 4.18, in Supplementary Material A. n = 260) scored higher than participants from the ICAR project. Yet, albeit significant ( t(306.45) = 4.66, p < 0.001, d = 0.21), effect sizes were smaller than the ones we retrieved Group Differences for the comparison with the Tel-SO within the current study. For the self-report-based classification, we found no dif- Pedo-SO/ + SO had higher scores on the ICAR test than Tel-SO, even ferences between Pedo-SO/ + SO and Tel-SO regarding when controlling for completion time via an ANCOVA design (F(1, 489) = 25.25, p < .001). Yet, note that although ANCOVAs represent a height, laterality index, and head injuries before and after common strategy to "control" for confounds, we have doubts that this age 13 (Table 5). Yet, we detected significant differences in strategy is appropriate and leads to meaningful conclusions (see Miller IQ scores, with Pedo-SO/ + SO achieving higher scores than & Chapman, 2001 for a non-technical discussion of common misuses Tel-SO. P-SO also emerged as significantly taller and more of ANCOVA in psychopathology research). 1 3 860 Archives of Sexual Behavior (2022) 51:849–866 Table 6 Planned comparisons (Helmert contrasts): Self-reported pedohebephilic men with vs. without convictions for sexual offending (Study 2) Variable Pedohebephilia, no sexual Pedohebephilia, sexual P-SO vs. P + SO offending (P –SO) offending (P + SO) M (SD) n M (SD) n t (df) p d Attraction to prepubescents 8.16 (2.41) 239 8.69 (1.92) 39 1.55 (59.48) .128 0.23 Attraction to early–mid-pubescents 7.98 (2.44) 239 8.03 (2.05) 39 0.12 (57.13) .908 0.02 ** b Attraction to adults 4.63 (2.53) 239 3.31 (2.81) 39 − 2.76 (48.56) .008 − 0.52 ** b Age 33.59 (13.14) 234 39.51 (10.84) 39 3.06 (58.31) .003 0.46 Viewing time score T1,2,3—T4,5 258 (860) 152 262 (659) 29 0.02 (487) .982 0.00 ** Height 179.38 (7.34) 224 175.77 (8.18) 35 − 2.72 (563) .007 − 0.49 EHI Laterality Index 0.66 (0.62) 239 0.79 (0.54) 39 1.27 (588) .167 0.21 *** b ICAR 10.35 (4.00) 224 6.78 (4.02) 36 − 4.95 (46.81) < .001 − 0.90 Head injuries before age 13 0.24 (0.61) 239 0.29 (0.90) 38 0.40 (587) .763 0.07 b c Head injuries after age 13 0.16 (0.56) 239 0.43 (0.99) 37 1.64 (39.63) .131 0.44 *** ** p < .001, p < .01 (two-sided) d = M –M / SD , calculated using the cohen.d function of the R package psych 1 2 pooled We used Welch’s correction due to unequal variances (as indicated by Levene test for equality of variances, center = median) p-value based on 1,000 bootstrap samples due to severe deviations from the assumption that residuals are normally distributed When the classification of sexual attraction was based on General Discussion VT, none of the contrasts for markers for neurodevelopmental differences reached significance (see Supplementary Mate - We could not corroborate our first hypothesis that pedo- rial Tables S3–S4). Note that sample sizes were smaller in hebephilic men from our two community samples differed the alternative classification, so failure to reach significance significantly from teleiophilic men with respect to height, was more likely. Yet, differences descriptively tended to be intelligence, non-right-handedness, or rate of head inju- similar in magnitude and direction, with the exception of ries during childhood. We did, however, find evidence for Pedo-SO/ + SO showing a stronger preference for the right the second hypothesis that pedohebephilic men who have hand compared to Tel-SO. sexually offended were smaller and less intelligent than non- offending pedohebephilic men. Replicating Gerwinn et al.'s Control Analyses (2019) findings, the current studies do not yield support for the theory that sexual interests in children among community To determine whether age could be a confound, we assessed men are linked to neurodevelopmental differences. Yet, in its link to IQ, height, and head injuries before and after age contrast to Gerwinn et al. (2019), which suffered from lacking 13 separately for each of the six tested subgroups (i.e., [1] statistical power, Study 2 was able to test differences between self-reported Pedo-SO, Pedo + SO, and Tel-SO; [2] VT- pedohebephilic and teleiophilic men at least 1-β = 0.80, with inferred Pedo-SO, Pedo + SO, and Tel-SO). For variables the exception of height. that deviated severely from the assumption of normality (in Because of the limitations of the present case–control this case head injuries before or after 13), we used boot- design and the impossibility to prove a null hypothesis in strapping procedures. For the self-report based classifica- classical hypothesis testing, our findings cannot conclusively tion, only 2/12 tests showed a significant link between older disprove the idea of a general link between indicators of neu- age and any of the dependent variables (head injuries before rodevelopmental perturbations and pedohebephilia. Never- age 13 and ICAR among Pedo-SO, r = 0.13, 95%CI based theless, the present data are positively commensurate with on bootstrap samples = [0.008, 0.29] and r = 0.14, p = 0.033, the notion that indicators of neurodevelopmental perturba- 95%CI = [0.01, 0.27], respectively). For the VT-based clas- tions such as, particularly, height and intelligence are linked sification, we could also only detect significant results for to (self-reported) offending status when pedophilic interest 1/12 tests (see Supplemental Material C). Considering the is kept constant in group comparisons. This dovetails with large number of tests and the direction of the effects, the the fact that all markers of neurodevelopmental perturba- evidence for age confounds was weak, as the number of sig- tions as assessed here are empirical correlates of criminal nificant tests does not fall far from the margin of what would behavior—as outlined in introduction (see also Beckley et al., be expected by chance at the given significance threshold of 2014; Bogaert, 2001; McKinlay et al., 2014; Moffitt et al., p < 0.05. 1994). Future research is needed to clarify if the importance 1 3 Archives of Sexual Behavior (2022) 51:849–866 861 of neurobiological alterations might have been overstated in self-referred pedohebephilic patients in treatment projects the previous literature, as part of the variance attributed to (Jahnke et al., 2015). pedohebephilia may in fact have to be attributed to other fac- The generalizability of the present findings is further con- tors due to the lack of specificity of the neurodevelopmental strained by the WEIRDness (Western, Educated, Industrial- perturbations hypothesis. ized, Rich, Democratic; Henrich et al., 2010) of the present samples. Note also that we cannot guarantee that Study 1 and Limitations and Outlook 2, without exception draw from two completely independent participant pools. It is possible that some participants are flu- Our results are subject to a number of caveats, some of which ent in both German and English, have visited both German are specific to our setting and methodology, while others and English language web forums for pedohebephilic people, are generally associated with research on proxy measures of and have participated in both studies. Future research draw- neurodevelopmental perturbations. First and foremost, cor- ing on online samples of pedohebephilic men should include relational analyses are unfit to prove causation—a fact that an item to identify and potentially screen out participants applies to most studies in the field of the neurodevelopmental from previous studies on the neurodevelopmental markers perturbations hypothesis and that increases the risk of fall- of pedohebephilia. ing victim to third variable confounds in case–control study Furthermore, both studies only included relatively small designs. numbers of pedohebephilic men who reported convictions for Additionally, our study populations are neither representa- sexual offenses. This means that even though some contrasts tive for teleiophilic nor pedohebephilic men. While clini- yielded significant results, small changes in participant char - cal, institutional, or forensic samples are subject to selection acteristics can lead to fluctuations in the results. Although biases as well, this means that results have to be interpreted difficult to attain, replications with larger samples of men carefully with respect to confounds and (self-selection) with sexual interests in children who have sexually offended biases. It is, for instance, possible that tall pedohebephilic are needed to corroborate these effects with more certainty. men with high levels of cognitive functioning were more Preferably, these should be men from the community to con- likely to participate than those with lower height or lower trol for selection factors related to being institutionalized in levels of cognitive abilities, given that they were aware of forensic settings. hypotheses from the academic literature. Yet, as online stud- Additionally, readers need to keep in mind that offending ies generally tend to oversample younger and more educated status was determined based on self-reported convictions for participants, this limitation also applies to the teleiophilic rape, child sexual abuse, and/or child pornography offenses groups. It is also possible that links between markers of neu- alone. Hence, there may be people in our non-offending rodevelopmental perturbations and pedohebephilia exist in sample who have sexually offended but were not detected or the population but cannot be identified when assessing sub- convicted, which makes it impossible to tell whether effects groups with high computer literacy. Even if pedohebephilia are a function of criminal sexual behavior or of being con- was linked to such markers, it may be impossible to detect victed for such acts. Previous research also indicates that men these links in samples without individuals with low levels convicted for child pornography offenses may differ from of cognitive functioning. Yet, note that in forensic settings, men convicted for child sexual offending on a number of pedohebephilic men who have been convicted for child por- relevant variables, such as antisociality (Babchishin et al., nography offending (i.e., indicating at least average levels 2015). Hence, it is possible that we would have found larger of computer literacy and cognitive capacity) showed more differences between the groups with and without a history markers of neurodevelopmental perturbations than teleio- of sexual offending, if we had had a higher rate of pedohebe- philic men who have committed the same crime (Blanchard philic men convicted for child sexual offending compared to et al., 2007). In the absence of representative samples of tel- child pornography offending. Yet, note that a study compar - eiophilic and pedophilic men, it is impossible to determine ing pedohebephilic people with either of these convictions if our results are due to such ceiling effects. shows more similarities than differences between the groups Moreover, participants from the pedohebephilia groups (Neutze et al., 2011). Furthermore, because we did not assess were recruited via forums addressing sexual attraction to chil- previous non-sexual convictions, we could not screen out dren, where they may have been looking for support because people with a significant non-sexual offense history as an they experienced increased levels of distress. Indeed, previ- indicator of antisociality. This is a potential limitation, as ous research indicates that pedohebephilic participants in antisociality may be linked to neurodevelopmental pertur- online surveys generally report higher levels of distress than bations. Yet, given the nature of our surveys and previous population-based or other non-clinical control samples. Yet, research among MTurk samples, our samples are unlikely to it is also shown that they are considerably less distressed than include people with marked antisocial traits. 1 3 862 Archives of Sexual Behavior (2022) 51:849–866 Another concern is that pedohebephilic participants might in its entirety, as the evidence for this hypothesis goes beyond have been more motivated to perform optimally in order to the factors examined here. better the image of their group. These motivational differ - Furthermore, with exception of the VT measure and the ences became apparent in the intelligence measure in Study 2, IQ test, the online environment forced us to rely completely which involved complex puzzles that required eo ff rt to solve. on self-report, which puts restraints on the validity of the data Teleiophilic MTurk workers, who purportedly had less to assessed. The tendency of men to over-report their height is gain from putting extra efforts into the cognitive tests, spent well-documented (Gorber et al., 2007), and pedophilic and much less time on solving the items than the people from teleiophilic men who have committed sexual offenses are no the pedohebephilic group. To get a perhaps more accurate exception to this rule: Fazio et al. (2014) report differences estimation of pedohebephilic men's cognitive abilities, we between actual and self-reported height varying between additionally compared responses from our pedohebephilic 2 and 4 cm, similar to differences observed in non-clinical sample in Study 2 with test data from Condon and Revelle's samples (e.g., Palta et al., 1982). This means that the data we (2014) validity study on the ICAR items. As participants in obtained on height in both our pedophilic and teleiophilic their dataset had been recruited online with the prospect to samples are likely to be biased. However, this is potentially a receive customized feedback, we expected that these partici- common limitation, at least for earlier studies, which, aston- pants had put more effort into solving the presented logical ishingly, either report to have relied on self-reported height puzzles compared to our teleiophilic sample. Yet, while these (e.g., Cantor et al., 2007) or had no access to information on participants scored higher on the test items than teleiophilic how height was assessed (Levenson & Ackerman, 2017; but men from Study 2, pedohebephilic men still emerged as more note that this is less likely to be the fact for newer research, intelligent. Fazio et al., 2017). Despite those problems, online research One of the main drawbacks of online research is the lack may represent the only feasible way of reaching participants of control that researchers can exert on the setting. In con- with pedohebephilia (as a dominant sexual attraction) outside trast to laboratory surveys, we were not able to make sure of clinical or forensic contexts. that participants stayed focused during the IQ test or cor- The present paper only focused on how men with pedo- rectly followed all instructions, and doubts remain whether hebephilic (who have and have not sexually offended) and IQ tests in non-controlled online environments compare teleiophilic interests differed in terms of neurodevelopmental with clinician administered IQ inventories. Yet, note that indicators. Future studies based on larger samples than the previous research found no structural differences regarding present one should consider conducting additional more fine- matrices intelligence scores obtained in online and offline grained analyses (e.g., to differentiate between pedophilic, samples (Ihme et al., 2009). Notably, although the presence hebephilic, and teleiophilic participants or between different of an interviewer can be advantageous, it may also introduce types of sexual offenses involving children). It also needs an expectancy bias on the interviewer’s side. For instance, to be stressed that, following conventions in the field, the clinical staff may look harder for anomalies when examining present study was conducted with a type II error rate of 0.20, pedophilic men, or they may more readily accept a teleio- meaning that there is a 20% chance that the null hypothesis philic man's self-report to have never had an accident leading (i.e., that there is no difference between pedohebephilic and to unconsciousness, prompting them to ask fewer follow-up teleiophilic men) is falsely accepted. questions. Reporting biases due to perceived demand effects on the interviewee’s side (and/or the larger clinical-forensic Conclusions context) are also conceivable. In the present studies, where pedophilic and non-pedophilic men were prompted with the Our findings underscore a general conclusiveness problem same set of questions, such types of bias are unlikely. arising from case–control designs: The interpretation of find- We were also limited in the range of variables we were ings is largely dependent on the composition of the comparison able to assess, as many potentially interesting outcome vari- groups. While there is no such thing as a representative sample ables are difficult (e.g., minor physical anomalies or 2D:4D of men with pedohebephilic interests, studies from community digit ratio; Jordan et al., 2020) or even impossible (e.g., brain settings represent an important corrective to data from pedo- morphology and functioning; Cantor et al., 2008; Massau hebephilic men "who are available for study because they are et al., 2017; Schiffer et al., 2017) to measure in an online either distressed by their sexual interests […] or criminally setting. For obvious reasons, this precludes the assessment charged for sexual offenses" (Seto, 2004, p. 323). Whatever of large numbers of pedohebephilic men without a history of sexual offending as a mostly hidden and hard-to-reach population. On the one hand, this limits our ability to judge The authors are willing to share the results of such more detailed the validity of the neurodevelopmental perturbations theory analyses based on our datasets (e.g., for the purpose of meta-analysis) upon request. 1 3 Archives of Sexual Behavior (2022) 51:849–866 863 processes may ultimately be involved in the development of Declarations pedohebephilic interests, evidence from recent surveys involv- Conflict of interest The authors declare that they have no conflict of ing non-offending pedohebephilic individuals indicates that interest. the claim that pedohebephilia is based on neurodevelopmen- tal perturbations might not be as inevitable and generalizable Consent to Participate Informed consent was obtained from all indi- as previously thought by bearing the potential to disentangle vidual participants included in the study. “institutional caseness,” criminality, and sexual interest (Ger- Consent to Publish Participants consented to the publication of their winn et al., 2018). Moreover, as the neurodevelopmental pertur- data in aggregated form. bation account is silent on the actual processes that are involved in developing pedohebephilia (and which thus have never been Ethical Approval This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the institutional subjected to empirical testing) only proxy measures that itself review board of the MSH Medical School for Study 1 and the institu- are just correlates of the supposed developmental trajectory tional review board of the Technische Universität Dresden for Study 2. have hitherto been examined. The potential of such correlates (i.e., the perturbation proxy measures used here) of correlates Open Access This article is licensed under a Creative Commons Attri- (i.e., one process among possibly many etiological pathways) to bution 4.0 International License, which permits use, sharing, adapta- be confounded with third variables is high and leads to substan- tion, distribution and reproduction in any medium or format, as long tial methodological problems in gathering empirical support as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes for the theory—specifically if it refers to a phenomenon that is were made. The images or other third party material in this article are linked to criminal behavior, publicly despised (Lehmann et al., included in the article's Creative Commons licence, unless indicated 2020), and has a low base-rate (Bártová et al., 2021; Dombert otherwise in a credit line to the material. If material is not included in et al., 2016). the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will Last but not least, it is noteworthy that data from two need to obtain permission directly from the copyright holder. To view a samples corroborated the validity of VT assessments in copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . self-identified community males with pedohebephilic sexual interests by yielding theoretically meaningful die ff rences in a known-groups approach. This can be regarded as the hitherto References most accurate estimator of the validity of VT measures, as prior studies rested on various group comparisons with men Abé, C., Adebahr, R., Liberg, B., Mannfolk, C., Lebedev, A., Eriks- convicted for child sexual abuse. The latter groups, however, son, J., Långström, N., & Rahm, C. (2021). Brain structure and clinical profile point to neurodevelopmental factors involved in are to a large degree only a pedohebephilia proxy as they are pedophilic disorder. Acta Psychiatrica Scandinavica, 143(4), composed of pedohebephilic and teleiophilic individuals thus 363–374. https:// doi. org/ 10. 1111/ acps. 13273 yielding only a conservative (i.e., lower-bound) estimator of Babchishin, K. M., Hanson, R. K., & VanZuylen, H. (2015). Online the magnitude of possible group differences (Schmidt et al., child pornography offenders are different: A meta-analysis of the characteristics of online and offline sex offenders against 2017). children. Archives of Sexual Behavior, 44(1), 45–66. https://doi. org/ 10. 1007/ s10508- 014- 0270-x Supplementary Information The online version contains supplemen- Bailey, A. A., & Hurd, P. L. (2005). Finger length ratio (2D:4D) cor- tary material available at https://doi. or g/10. 1007/ s10508- 021- 02228-w . relates with physical aggression in men but not in women. Bio- logical Psychology, 68(3), 215–222. https:// doi. org/ 10. 1016/j. Acknowledgements We are grateful to David M. Condon for sharing biops ycho. 2004. 05. 001 his expertise on online intelligence assessment and to Ian V. McPhail Bártová, K., Androvičová, R., Krejčová, L., Weiss, P., & Klapilová, for commenting on an earlier draft of this research. K. (2021). The prevalence of paraphilic interests in the Czech population: Preference, arousal, the use of pornography, fantasy, Author Contributions All authors developed the study concept and and behavior. Journal of Sex Research, 58, 86–96. https:// doi. design for Study 1. Data collection for Study 1 was performed by A.K. org/ 10. 1080/ 00224 499. 2019. 17074 68 under the supervision of J.H. and S.J. Data analysis for Study 1 was Becerra García, J. A. (2009). Etiology of pedophilia from a neurode- conducted by S.J. and A.S. S.J. and A.S. developed the study concept velopmental perspective: Markers and brain alterations. Revista and collected and analyzed the data for Study 2. S.J. drafted the paper, de Psiquiatría y Salud Mental, 2(4), 190–196. https:// doi. org/ and A.S. and J.H. provided critical revisions. All authors approved the 10. 1016/ S2173- 5050(09) 70051-2 final version of the paper for submission. Beckley, A. L., Kuja-Halkola, R., Lundholm, L., Långström, N., & Frisell, T. (2014). Association of height and violent criminality: Results from a Swedish total population study. International Funding Open access funding provided by University of Bergen Journal of Epidemiology, 43(3), 835–842. h t t p s : / / d o i . o r g / 1 0 . (incl Haukeland University Hospital). This study received no specific 1093/ ije/ dyt274 funding. 1 3 864 Archives of Sexual Behavior (2022) 51:849–866 Blanchard, R., Christensen, B. K., Strong, S. M., Cantor, J. M., sexual interest in prepubescent children? Journal of Sex Research, Kuban, M. E., Klassen, P., Dickey, R., & Blak, T. (2002). Ret- 53(2), 214–223. https://doi. or g/10. 1080/ 00224 499. 2015. 10201 08 rospective self-reports of childhood accidents causing uncon- Dyshniku, F., Murray, M. E., Fazio, R. L., Lykins, A. D., & Cantor, J. sciousness in phallometrically diagnosed pedophiles. Archives M. (2015). Minor physical anomalies as a window into the pre- of Sexual Behavior, 31(6), 511–526. https:// doi. org/ 10. 1023/A: natal origins of pedophilia. Archives of Sexual Behavior, 44(8), 10206 59331 965 2151–2159. https:// doi. org/ 10. 1007/ s10508- 015- 0564-7 Blanchard, R., Kolla, N. J., Cantor, J. M., Klassen, P. E., Dickey, R., Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: Kuban, M. E., & Blak, T. (2007). IQ, handedness, and pedophilia A flexible statistical power analysis program for the social, behav - in adult male patients stratified by referral source. Sexual Abuse: ioral, and biomedical sciences. Behavior Research Methods, 39(2), A Journal of Research and Treatment, 19(3), 285–309. https://doi. 175–191. org/ 10. 1007/ s11194- 007- 9049-0 Fazio, R. L. (2018). Toward a neurodevelopmental understanding of Blanchard, R., Kuban, M. E., Klassen, P., Dickey, R., Christensen, B. K., pedophilia. Journal of Sexual Medicine, 15(9), 1205–1207. https:// Cantor, J. M., & Blak, T. (2003). Self-reported head injuries before doi. org/ 10. 1016/j. jsxm. 2018. 04. 631 and after age 13 in pedophilic and nonpedophilic men referred for Fazio, R. L., Dyshniku, F., Lykins, A. D., & Cantor, J. M. (2017). Leg clinical assessment. Archives of Sexual Behavior, 32(6), 573–581. length versus torso length in pedophilia: Further evidence of https:// doi. org/ 10. 1023/A: 10260 93612 434 atypical physical development early in life. Sexual Abuse, 29(5), Bogaert, A. F. (2001). Handedness, criminality, and sexual offend- 500–514. https:// doi. org/ 10. 1177/ 10790 63215 609936 ing. Neuropsychologia, 39(5), 465–469. https:// doi. org/ 10. 1016/ Fazio, R. L., Lykins, A. D., & Cantor, J. M. (2014). Elevated rates S0028- 3932(00) 00134-2 of atypical handedness in paedophilia: Theory and implications. Boscarino, J. A., & Hoffman, S. N. (2007). Consistent association Laterality, 19(6), 690–704. https:// doi. org/ 10. 1080/ 13576 50X. between mixed lateral preference and PTSD: Confirmation among 2014. 898648 a national study of 2490 US Army Vietnam veterans. Psychoso- Fink, B., Neave, N., Laughton, K., & Manning, J. T. (2006). Second matic Medicine, 69(4), 365–369. https:// doi. or g/ 10. 1097/ PSY . to fourth digit ratio and sensation seeking. Personality and Indi- 0b013 e3180 5fe2bc vidual Differences, 41(7), 1253–1262. https:// doi. org/ 10. 1016/j. Brower, M. C., & Price, B. H. (2001). Neuropsychiatry of frontal lobe paid. 2006. 05. 002 dysfunction in violent and criminal behaviour: A critical review. Gerwinn, H., Weiß, S., Tenbergen, G., Amelung, T., Födisch, C., Pohl, Journal of Neurology, Neurosurgery & Psychiatry, 71(6), 720– A., Massau, C., Kneer, J., Mohnke, S., Kärgel, C., Wittfoth, M., 726. https:// doi. org/ 10. 1136/ jnnp. 71.6. 720 Jung, S., Drumkova, K., Schiltz, K., Walter, M., Beier, K. M., Cantor, J. M., Blanchard, R., Christensen, B. K., Dickey, R., Klassen, Walter, H., Ponseti, J., Schiffer, B., & Kruger, T. H. C. (2018). P. E., Beckstead, A. L., Blak, T., & Kuban, M. E. (2004). Intelli- Clinical characteristics associated with paedophilia and child sex gence, memory, and handedness in pedophilia. Neuropsychology, offending—Differentiating sexual preference from offence status. 18(1), 3–14. European Psychiatry, 51, 74–85. https://doi. or g/10. 1016/j. eur psy. Cantor, J. M., Blanchard, R., Robichaud, L. K., & Christensen, B. K. 2018. 02. 002 (2005a). Quantitative reanalysis of aggregate data on IQ in sexual Gorber, S. C., Tremblay, M., Moher, D., & Gorber, B. (2007). A compar- offenders. Psychological Bulletin, 131(4), 555–568. https:// doi. ison of direct vs. self-report measures for assessing height, weight org/ 10. 1037/ 0033- 2909. 131.4. 555 and body mass index: A systematic review. Obesity Reviews, 8(4), Cantor, J. M., Kabani, N., Christensen, B. K., Zipursky, R. B., Barbaree, 307–326. https:// doi. org/ 10. 1111/j. 1467- 789X. 2007. 00347.x H. E., Dickey, R., Klassen, P. E., Mikulis, D. J., Kuban, M. E., Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest peo- Blak, T., Richards, B. A., Hanratty, M. K., & Blanchard, R. (2008). ple in the world? Behavioral and Brain Sciences, 33(2–3), 61–83. Cerebral white matter deficiencies in pedophilic men. Journal of https:// doi. org/ 10. 1017/ S0140 525X0 99915 2X Psychiatric Research, 42(3), 167–183. https:// doi. org/ 10. 1016/j. Hubert, M., & Vandervieren, E. (2008). An adjusted boxplot for skewed jpsyc hires. 2007. 10. 013 distributions. Computational Statistics & Data Analysis, 52(12), Cantor, J. M., Klassen, P. E., Dickey, R., Christensen, B. K., Kuban, M. 5186–5201. https:// doi. org/ 10. 1016/j. csda. 2007. 11. 008 E., Blak, T., Williams, N. S., & Blanchard, R. (2005b). Handedness Ihme, J. M., Lemke, F., Lieder, K., Martin, F., Müller, J. C., & Schmidt, in pedophilia and hebephilia. Archives of Sexual Behavior, 34(4), S. (2009). Comparison of ability tests administered online and in 447–459. https:// doi. org/ 10. 1007/ s10508- 005- 4344-7 the laboratory. Behavior Research Methods, 41(4), 1183–1189. Cantor, J. M., Kuban, M. E., Blak, T., Klassen, P. E., Dickey, R., & https:// doi. org/ 10. 3758/ BRM. 41.4. 1183 Blanchard, R. (2007). Physical height in pedophilic and hebephilic Imhoff, R., Schmidt, A. F., Nordsiek, U., Luzar, C., Young, A. W., sexual offenders. Sexual Abuse, 19(4), 395–407. https://doi. or g/10. & Banse, R. (2010). Viewing Time effects revisited: Prolonged 1177/ 10790 63207 01900 405 response latencies for sexually attractive targets under restricted Condon, D. M., & Revelle, W. (2014). The international cognitive abil- task conditions. Archives of Sexual Behavior, 39(6), 1275–1288. ity resource: Development and initial validation of a public-domain https:// doi. org/ 10. 1007/ s10508- 009- 9595-2 measure. Intelligence, 43, 52–64. https:// doi. org/ 10. 1016/j. intell. Jahnke, S., & Malón, A. (2019). How pedohebephilic men think about 2014. 01. 004 adult-child sex: Effects of child gender and physical maturity. Psy- Condon, D. M., & Revelle, W. (2015). Selected ICAR data from the chology, Crime & Law, 25(1), 90–107. https:// doi. org/ 10. 1080/ SAPA-Project: Development and initial validation of a public-10683 16X. 2018. 15036 65 domain measure [Data set]. https:// doi. or g/ 10. 7910/ D VN/ Jahnke, S., Schmidt, A. F., Geradt, M., & Hoyer, J. (2015). Stigma- AD9RVY related stress and its correlates among men with pedophilic sexual Denny, K. (2009). Handedness and depression: Evidence from a large interests. Archives of Sexual Behavior, 44(8), 2173–2187. https:// population survey. Laterality, 14(3), 246–255. https:// doi. org/ 10. doi. org/ 10. 1007/ s10508- 015- 0503-7 1080/ 13576 50080 23628 69 Jahnke, S., Schmidt, A. F., & Hoyer, J. (2021). Perceived non-coercive Denny, K. (2011). Handedness and drinking behaviour. British Journal childhood sexual experiences among community men with and of Health Psychology, 16(2), 386–395. https:// doi. org/ 10. 1348/ without pedohebephilia in two non-matched case-control stud- 13591 0710X 515705 ies [Manuscript submitted for publication]. Department of Health Promotion and Development, University of Bergen. Dombert, B., Schmidt, A. F., Banse, R., Briken, P., Hoyer, J., Neutze, J., & Osterheider, M. (2016). How common is men’s self-reported 1 3 Archives of Sexual Behavior (2022) 51:849–866 865 Jordan, K., Wild, T. S. N., Fromberger, P., Müller, I., & Müller, J. L. abuse in a community sample of pedophiles and hebephiles. Sex- (2020). Are there any biomarkers for pedophilia and sexual child ual Abuse, 23(2), 212–242. https:// doi. org/ 10. 1177/ 10790 63210 abuse? A review. Frontiers in Psychiatry, 10, 940. https:// doi. org/ 382043 10. 3389/ fpsyt. 2019. 00940 Ogunfowokan, A. A., Olagunju, O. E., Olajubu, A. O., Faremi, F. A., Joyal, C. C., Black, D. N., & Dassylva, B. (2007). The neuropsychology Oloyede, A. S., & Sharps, P. W. (2016). Correlates of self-report of and neurology of sexual deviance: A review and pilot study. Sexual rape among male school adolescents in Ile-Ife, Nigeria. Journal of Abuse: A Journal of Research and Treatment, 19(2), 155–173. Interpersonal Violence, 31(4), 555–571. https:// doi. org/10 .11 77/ https:// doi. org/ 10. 1007/ s11194- 007- 9045-408862 60514 556104 Joyal, C. C., & Carpentier, J. (2021). Concordance and discordance Palta, M., Prineas, R. J., Berman, R., & Hannan, P. (1982). Comparison between paraphilic interests and behaviors: A follow-up study. of self-reported and measured height and weight. American Jour- Journal of Sex Research. https://doi. or g/10. 1080/ 00224 499. 2021. nal of Epidemiology, 115(2), 223–230. https:// doi. org/ 10. 1093/ 19868 01oxfor djour nals. aje. a1132 94 Joyal, C. C., Kärgel, C., Kneer, J., Amelung, T., Mohnke, S., Tenbergen, Pedneault, C. I., Hilgard, J., Pettersen, C., Hermann, C. A., White, K., G., Walter, H., & Kruger, T. H. C. (2019). The neurobiological & Nunes, K. L. (2021). How well do indirect measures assess origins of pedophilia: Not that simple. Journal of Sexual Medicine, sexual interest in children? A meta-analysis. Journal of Consult- 16(1), 153–154. https:// doi. org/ 10. 1016/j. jsxm. 2018. 10. 015 ing and Clinical Psychology, 89(4), 350–363. https:// doi. org/ 10. Lalumière, M. L., Blanchard, R., & Zucker, K. J. (2000). Sexual ori-1037/ ccp00 00627 entation and handedness in men and women: A meta-analysis. Poeppl, T. B., Eickhoff, S. B., Fox, P. T., Laird, A. R., Rupprecht, R., Psychological Bulletin, 126(4), 575–592. https://doi. or g/10. 1037/ Langguth, B., & Bzdok, D. (2015). Connectivity and functional 0033- 2909. 126.4. 575 profiling of abnormal brain structures in pedophilia. Human Brain Lehmann, R. J. B., Schmidt, A. F., & Jahnke, S. (2020). Stigmatiza- Mapping, 36(6), 2374–2386. https:// doi. org/ 10. 1002/ hbm. 22777 tion of paraphilias and psychological conditions linked to sexual Poeppl, T. B., Nitschke, J., Dombert, B., Santtila, P., Greenlee, M. W., offending. Journal of Sex Research, 58(4), 438–447. https:// doi. Osterheider, M., & Mokros, A. (2011). Functional cortical and org/ 10. 1080/ 00224 499. 2020. 17547 48 subcortical abnormalities in pedophilia: A combined study using Levenson, J. S., & Ackerman, A. R. (2017). The relationship between a choice reaction time task and fMRI. Journal of Sexual Medi- sex offender height and pedophilic interest. Deviant Behavior, 38, cine, 8(6), 1660–1674. https://doi. o rg/10. 1111/j. 1743- 610 9.2011. 1383–1392. https:// doi. org/ 10. 1080/ 01639 625. 2016. 12549 86 02248.x Lichtenberger, E. O., & Kaufman, A. S. (2009). Essentials of WAIS-IV Ray, J. J. (1984). The reliability of short social desirability scales. Jour- assessment. John Wiley & Sons. nal of Social Psychology, 123(1), 133. Liu, J. (2011). Early health risk factors for violence: Conceptualiza- Schiffer, B., Amelung, T., Pohl, A., Kaergel, C., Tenbergen, G., Ger - tion, evidence, and implications. Aggression and Violent Behavior, winn, H., Mohnke, S., Massau, C., Matthias, W., Weiß, S., Marr, 16(1), 63–73. https:// doi. org/ 10. 1016/j. avb. 2010. 12. 003 V., Beier, K. M., Walter, M., Ponseti, J., Krüger, T. H. C., Schiltz, Maechler, M., Rousseeuw, P., Croux, C., Todorov, V., Ruckstuhl, A., K., & Walter, H. (2017). Gray matter anomalies in pedophiles Salibian-Barrera, M., Verbeke, T., Koller, M., Conceicao, E. L. with and without a history of child sexual offending. Translational T., & di Palma, M. A. (2019). robustbase: Basic robust statistics Psychiatry, 7(5), e1129. https:// doi. org/ 10. 1038/ tp. 2017. 96 (0.93–5) [Computer software]. https://CRAN .R-pr oject. or g/pac ka Schiffer, B., Peschel, T., Paul, T., Gizewski, E., Forsting, M., Leygraf, ge= robus tbase N., Schedlowski, M., & Krueger, T. H. C. (2007). Structural brain Massau, C., Tenbergen, G., Kärgel, C., Weiß, S., Gerwinn, H., Pohl, abnormalities in the frontostriatal system and cerebellum in pedo- A., Amelung, T., Mohnke, S., Kneer, J., Wittfoth, M., Ristow, I., philia. Journal of Psychiatric Research, 41(9), 753–762. https:// Schiltz, K., Beier, K. M., Ponseti, J., Walter, M., Kruger, T. H. doi. org/ 10. 1016/j. jpsyc hires. 2006. 06. 003 C., Walter, H., & Schiffer, B. (2017). Executive functioning in Schipolowski, S., Wilhelm, O., Schroeders, U., Kovaleva, A., Kemper, pedophilia and child sexual offending. Journal of the International C. J., & Rammstedt, B. (2014). Eine kurze Skala zur Messung- Neuropsychological Society, 23(6), 460–470. https:// doi. org/ 10. kristalliner Intelligenz: Die Kurzskala gc des Berliner Tests zur 1017/ S1355 61771 70003 15 Erfassung Fluider und Kristalliner Intelligenz (BEFKI GC-K). McKinlay, A., Grace, R. C., McLellan, T., Roger, D., Clarbour, J., & GESIS-Working Papers 2014/29. http://nbn- r esolving. de/ ur n:nbn: MacFarlane, M. R. (2014). Predicting adult offending behavior for de: 0168- ssoar- 377864 individuals who experienced a traumatic brain injury during child- Schmidt, A. F., Babchishin, K. M., & Lehmann, R. J. B. (2017). A meta- hood. Journal of Head Trauma Rehabilitation, 29(6), 507–513. analysis of viewing time measures of sexual interest in children. https:// doi. org/ 10. 1097/ HTR. 00000 00000 000000 Archives of Sexual Behavior, 46(1), 287–300. https:// doi. org/ 10. McPhail, I. V., & Cantor, J. M. (2015). Pedophilia, height, and the 1007/ s10508- 016- 0806-3 magnitude of the association: A research note. Deviant Behavior, Schmidt, A. F., Mokros, A., & Banse, R. (2013). Is pedophilic sexual 36(4), 288–292. preference continuous? A taxometric analysis based on direct and Merz, Z. C., Lace, J. W., & Eisenstein, A. M. (2020). Examining broad indirect measures. Psychological Assessment, 25(4), 1146–1153. intellectual abilities obtained within an mTurk internet sample. https:// doi. org/ 10. 1037/ a0033 326 Current Psychology. https://doi. or g/10. 1007/ s12144- 020- 00741-0 Seto, M. C. (2008). Pedophilia and sexual offending against children: Miller, G. A., & Chapman, J. P. (2001). Misunderstanding analysis Theory, assessment, and intervention (1st ed.). American Psycho- of covariance. Journal of Abnormal Psychology, 110(1), 40–48. logical Association. https:// doi. org/ 10. 1037/ 11639- 000. https:// doi. org/ 10. 1037// 0021- 843x. 110.1. 40 Seto, M. C. (2004). Pedophilia and sexual offenses against children. Moffitt, T. E., Lynam, D. R., & Silva, P. A. (1994). Neuropsychological Annual Review of Sex Research, 15(1), 321–361. https:// doi. org/ tests predicting persistent male delinquency. Criminology, 32(2), 10. 1080/ 10532 528. 2004. 10559 823 277–300. https:// doi. org/ 10. 1111/j. 1745- 9125. 1994. tb011 55.x Seto, M. C. (2012). Is pedophilia a sexual orientation? Archives Mokros, A., Habermeyer, E., & Küchenhoff, H. (2018). The uncertainty of Sexual Behavior, 41(1), 231–236. https:// doi. or g/ 10. 1007/ of psychological and psychiatric diagnoses. Psychological Assess-s10508- 011- 9882-6 ment, 30(4), 556–560. https:// doi. org/ 10. 1037/ pas00 00524 Singy, P. (2015). Hebephilia: A postmortem dissection. Archives of Sexual Behavior, 44(5), 1109–1116. https:// doi. or g/ 10. 1007/ Neutze, J., Seto, M. C., Schaefer, G. A., Mundt, I. A., & Beier, K. M. s10508- 015- 0542-0 (2011). Predictors of child pornography offenses and child sexual 1 3 866 Archives of Sexual Behavior (2022) 51:849–866 Stephens, S., Seto, M. C., Goodwill, A. M., & Cantor, J. M. (2017). Welsch, R., Schmidt, A. F., Turner, D., & Rettenberger, M. (2021). Test– Evidence of construct validity in the assessment of hebephilia. retest reliability and temporal agreement of direct and indirect Archives of Sexual Behavior, 46(1), 301–309. https:// doi. org/ 10. sexual interest measures. Sexual Abuse, 33(3), 339–360. https:// 1007/ s10508- 016- 0907-zdoi. org/ 10. 1177/ 10790 63220 904354 Tanner, J. M. (1990). Foetus into man: Physical growth from conception Young, S. R., & Keith, T. Z. (2020). An examination of the convergent to maturity. Harvard University Press. validity of the ICAR16 and WAIS-IV. Journal of Psychoeduca- Tenbergen, G., Wittfoth, M., Frieling, H., Ponseti, J., Walter, M., Wal- tional Assessment, 38(8), 1052–1059. https:// doi. or g/ 10. 1177/ ter, H., Beier, K. M., Schiffer, B., & Kruger, T. H. C. (2015). The 07342 82920 943455 neurobiology and psychology of pedophilia: Recent advances and Yule, M. A., Brotto, L. A., & Gorzalka, B. B. (2014). Biological markers challenges. Frontiers in Human Neuroscience. https:// doi. org/ 10. of asexuality: Handedness, birth order, and finger length ratios in 3389/ fnhum. 2015. 00344 self-identified asexual men and women. Archives of Sexual Behav - Tenconi, E., Santonastaso, P., Degortes, D., Bosello, R., Titton, F., ior, 43(2), 299–310. https:// doi. org/ 10. 1007/ s10508- 013- 0175-0 Mapelli, D., & Favaro, A. (2010). Set-shifting abilities, central coherence, and handedness in anorexia nervosa patients, their unaf- Publisher's Note Springer Nature remains neutral with regard to fected siblings and healthy controls: Exploring putative endophe- jurisdictional claims in published maps and institutional affiliations. notypes. World Journal of Biological Psychiatry, 11(6), 813–823. https:// doi. org/ 10. 3109/ 15622 975. 2010. 483250 Pacific Psychological Assessment Cooperation. (2004). The Not-Real People stimulus set for assessment of sexual interest. Author. 1 3

Journal

Archives of Sexual BehaviorSpringer Journals

Published: Feb 1, 2022

Keywords: Pedophilia; Child sexual abuse; Neurobiology; Neurodevelopment; Etiology; DSM-5

There are no references for this article.