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

Learn More →

Latency-Based and Psychophysiological Measures of Sexual Interest Show Convergent and Concurrent Validity

Latency-Based and Psychophysiological Measures of Sexual Interest Show Convergent and Concurrent... Arch Sex Behav (2018) 47:637–649 https://doi.org/10.1007/s10508-017-1133-z OR IG INA L PAP E R Latency-BasedandPsychophysiologicalMeasuresofSexualInterest Show Convergent and Concurrent Validity 1 2 1 • • Caoilte O Ciardha Janice Attard-Johnson Markus Bindemann Received: 9 January 2017 / Revised: 27 November 2017 / Accepted: 2 December 2017 / Published online: 20 December 2017 The Author(s) 2017. This article is an open access publication Abstract Latency-based measures of sexual interest require are available to researchers for this purpose (Savin-Williams & additional evidence of validity, as do newer pupil dilation approaches. Vrangalova, 2013). However, for research examining factors A total of 102 community men completed six latency-based mea- such as the cognitive mechanisms of sexual interests (e.g., sures of sexual interest. Pupillary responses were recorded during Ortigue & Bianchi-Demicheli, 2008), it becomes important three of these tasks and in an additional task where no participant to be able to measure those interests less directly, or even implic- response was required. For adult stimuli, there was a high degree itly. Additionally, there are applied contexts where understand- of intercorrelation between measures, suggesting that tasks may ing individuals’ sexual interests is clinically important but poten- be measuring the same underlying construct (convergent valid- tially vulnerable to social desirability biases—for example in the ity). In addition to being correlated with one another, mea- case of individuals who may have committed sexual offenses sures also predicted participants’ self-reported sexual inter- (see Kalmus & Beech, 2005). For this reason, much of the basic est, demonstrating concurrent validity (i.e., the ability of a task research that has examined the validity of indirect measures of to predict a more validated, simultaneously recorded, measure). sexual interest has done so with a distal or proximal goal of Latency-based and pupillometric approaches also showed pre- improving measurement of sexual interest relevant to forensic liminary evidence of concurrent validity in predicting both self- contexts, especially relating to sexual offending against chil- reported interest in child molestation and viewing pornographic dren. In the current study, we examined the validity of multiple material containing children. Taken together, the study findings methods of assessing sexual interest among male participants, build on the evidence base for the validity of latency-based again with the distal goal of improving forensic assessments, and pupillometric measures of sexual interest. but with relevance for broader sexology research. Despite perceptions to the contrary, not all individuals who Keywords Sexual interest  Viewing time  Reaction time  commit child sexual abuse have a sexual interest in children Pupil dilation  Pedophilia (Seto, 2008). While it may not be a universal treatment need, sexual interest in children is a leading predictor of sexual reof- fending among those convicted of a sexual offense (Hanson & Introduction Morton-Bourgon, 2004). Accurately assessing sexual interest is, therefore, vital for targeting criminogenic treatment needs For many research topics relating to sexuality and sexual behav- within rehabilitation and for estimating and managing risk. ior, it is sufficient for participants to self-report their sexual ori- The current ‘gold standard’ in assessing sexual interest is by mea- entation or sexual interests. A large number of self-report tools suring genital responses, typically using penile plethysmography (PPG). A recent meta-analysis by McPhail et al. (2017) found & Caoilte O Ciardha that PPG is a valid measure of pedohebephilic interest, but that C.C.OCiardha@kent.ac.uk validity varies depending on PPG methodology. Historically, PPG has faced questions of validity, reliability, and admissi- School of Psychology, Keynes College, University of Kent, Canterbury CT2 7NP, UK bility in court (for a review, see Laws, 2009). PPG is not rou- tinely used, for example, by Her Majesty’s Prison and Proba- Department of Psychology, Bournemouth University, Poole, tion Service in the UK in the assessment of individuals who UK 123 638 Arch Sex Behav (2018) 47:637–649 have committed sexual offenses. Where genital arousal is used While the IAT hinges on a facilitation effect, other reaction in sex research more broadly, there is a risk to the representa- time tasks depend on stimuli interfering with the primary task tiveness and generalizability of findings due to volunteer bias demands, thus suggesting that they hold some salience for the (Morokoff, 1986; Wolchik, Braver, & Jensen, 1985). Practi- participant. A variant of the choice reaction time (CRT) task tioners and researchers carrying out assessments of sexual inter- has been used where participants are instructed to report the est, therefore, need other indicators of sexual interest to comple- location of a dot on a screen while ignoring potentially sex- ment or provide an alternative to the use of PPG. ually appetitive stimuli on which that dot is overlaid. The As a response to the need for additional valid measures of CRT task typically results in longer reaction times to stimuli sexual interest, researchers have tested response latency-based consistent with participants’ sexual interests (e.g., Mokros, tasks (see Schmidt, Banse, & Imhoff, 2015; Thornton & Laws, Dombert, Osterheider, Zappala `, & Santtila, 2010; Wright & 2009). The most established task uses a viewing time paradigm Adams, 1994). Similarly, the modified Stroop task asks par- (Abel, Huffman, Warberg, & Holland, 1998) and is commer- ticipants to name the color of stimuli while ignoring their con- cially available as a technique for measuring sexual interest tent. Pictorial (O Ciardha & Gormley, 2012) and word (Price (for a meta-analysis of viewing time research, see Schmidt, & Hanson, 2007; Smith & Waterman, 2004) versions of the Babchishin, & Lehmann, 2017). Viewing time is based on the modified Stroop task have shown an interference effect of observation that individuals will spend longer looking at images stimulus content when that content is sexually salient. The they find sexually appealing than control or unappealing images. current study focuses on viewing time, IAT, CRT, and pictorial The task typically asks participants to rate images for attractive- modified Stroop (P-MST) tasks as latency-based measures of ness while simultaneously measuring response latency. Slowing sexual interest (for a broader review, see Schmidt et al., 2015). down of response times is interpreted as indicative of an indi- Compared to men, women—especially heterosexual women— vidual’s sexual interests. While the task is typically understood do not demonstrate the same level of category-specific respond- in relation to Singer’s (1984) theory of an aesthetic first phase in ing in physiological or latency-based tasks measuring sexual sexual arousal, the mechanisms underpinning this relatively interest (Snowden & Gray, 2013). For this reason, combined robust phenomenon are currently poorly understood (Imhoff with the lower prevalence rate of female-perpetrated sexual et al., 2010). There are concerns about the degree to which view- abuse, we focus only on the validity of tasks measuring sexual ing time tasks may be open to manipulation. Schmidt et al. (2017) interest among men. expressed surprise that there is a lack of research on the fakeability No single latency-based task currently holds a sufficient of viewing time. Additionally, for those undertaking assessment weight of evidence regarding its validity and reliability to be in adversarial settings, detailed task descriptions are available fully acceptable as a clinically useful measure of sexual inter- online; therefore, viewing time alone is not likely to provide a est. Indeed, case formulation is likely to be more reliably informed truly reliable measure of sexual interest for forensic assessment. by a confluence of clinically relevant information from multiple Other latency-based measures of sexual interest typically assessment methods rather than by identifying the single best demand that the participant carries out a task quickly and accu- performing task (Schmidt et al., 2017). Therefore, the question rately, while being presented with stimuli that may be related to of convergent and predictive/concurrent validity is of primary their sexual interests. Due to the fact that a rapid response is importance in building an evidence base for the clinical use of required in these tasks, we use the term reaction time tasks to measures of sexual interest. refer to them collectively and to differentiate them from view- Several studies have, to date, examined the convergent valid- ing time (Maletzky, 2003). In these tasks, sexually salient stimuli ity of different latency-based measures relating to sexual inter- may hinder or facilitate task performance in a way that is indica- est. Meta-analysis (Schmidt et al., 2017) based on a handful of tive of participants’ sexual interests. The Implicit Association studies has shown associations with small to medium effect Test (IAT) applied to the measurement of sexual interest asks sizes between viewing time and self-report, IAT, PPG, and the participants to categorize stimuli as quickly as they can use Screening Scale for Pedophilic Interests (Seto & Lalumie `re, 2001). two buttons (for a meta-analysis, see Babchishin, Nunes, & An earlier meta-analysis examining IAT research (Babchishin Hermann, 2013). In the trials of interest, each button has a et al., 2013) reported relationships with moderate effect sizes concept category and an attribute category assigned to it. If with viewing time, self-report, and the Screening Scale for the participant holds an association between the paired cat- Pedophilic Interests. egories, they should experience a facilitation effect when Ro ¨ nspies et al. (2015) compared viewing time with CRT, required to categorize stimuli. For example, Nunes, Firestone, and an adaptation of the Implicit Relational Assessment Pro- and Baldwin (2007) found that, when compared with other cedure (Barnes-Holmes et al., 2006) in measuring sexual ori- inmates, individuals who had committed sexual offenses against entation among male participants with exclusive sexual inter- children were quicker to categorize stimuli when paired cate- ests. The latter task is similar to the IAT but assesses an abso- gories were child/sexy and adult/not sexy than when the pairs lute rather than relative strength of relatedness/association were in the opposite arrangement. between concepts. Ro ¨ nspies et al. reported a convergence bet- 123 Arch Sex Behav (2018) 47:637–649 639 ween viewing time and the Implicit Relational Assessment Pro- respondence between pupillary responses and genital responses, cedure, both of which correlated with sexual orientation. The CRT particularly in males. Using non-nude images rather than videos, task used showed little evidence of construct validity. This was Attard-Johnson et al. (2016) found that straight men without a consistent with the findings of O Ciardha and Gormley (2013) self-reported sexual interest in children showed dilation to adult but not with the bulk of the published research which appears females but not to images of adult males or children. The current to demonstrate that the CRT can be used as a valid measure of study is the first to examine pupil dilation alongside multiple age-appropriate sexual interest or sexual interest in children latency-based measures of sexual interest. (Dombert et al., 2015; Gress, Anderson, & Laws, 2011;Mokros The study had two aims. First, we examined the convergent et al., 2010; Santtila et al., 2009;Wright & Adams, 1994). validity of viewing time, reaction time, and pupillometric mea- Using a community sample, the current study examined the sures of sexual interest in a non-offender sample. Participants convergent validity of viewing time (using three different sets performed a free-viewing task and three viewing time tasks (with of stimuli) alongside three reaction time tasks: a CRT task, a different stimuli) whereby pupil size was recorded, as well as P-MST, and a variant of the IAT. All tasks included both adult a CRT, a P-MST, and an IAT variant. The IAT variant (which and child stimuli. We tested concurrent validity by examining we refer to as a factorial IAT) included adult and child stimuli the relationship between all tasks and self-reported sexual inter- representing the concept categories of male and female.We ests in adults and some self-report indicators of possible sexual hypothesized that all measures would covary with one another, interest in children. No previous study has simultaneously com- demonstrating convergent validity, and that they would predict pared as many measures. Importantly, our study did not solely self-reported sexual interest, demonstrating concurrent valid- examine latency and self-report-based measures. We also incor ity. Second, we examined whether these tasks could also predict porated a psychophysiological measure of sexual interest into sexual interest toward children in this sample, further demon- the battery of tasks completed by participants. strating concurrent validity specifically in terms of their foren- The processes that facilitate or hinder speedy task perfor- sic application. This second aim was more exploratory given mance in reaction time tasks are typically seen as relatively the expectation of a low rate of sexual interest in children in a automatic and, therefore, less vulnerable to faking than other non-offender population. Santtila et al. (2015), for example, approaches. Where faking occurs, there is evidence that it is reported a one-year incidence of 3.3% for self-reported sexual detectable and may be correctable (Cvencek, Greenwald, Brown, interest in children among adult men. Gray, & Snowden, 2010). However, latency-based measures require input from the participant in the form of button-presses or vocal responding, which presents an opportunity for noise Method and/or deliberate manipulation to affect results. We, therefore, also included a psychophysiological method of assessing sex- Participants ual interest that may be more automatic than tasks requiring a response output from participants. Viewing content that is sex- One hundred and two men participated in this study in return ually appetitive elicits an instant automatic dilation of an for a small payment or course credit. Asamplesizetargetof100 observer’s pupils consistent with their sexual orientation (Attard- participants was set based on a combination of experience with Johnson, Bindemann, & O Ciardha, 2016; Rieger et al., 2015; previous experiments, pragmatic considerations (e.g., available Rieger & Savin-Williams, 2012). Importantly, pupil dilation funding), and a priori power analysis. We calculated the required appears highly resilient to deliberate manipulation (Laeng, sample size to test an overall multiple linear regression model Sirois, & Gredeba ¨ck, 2012). with ten predictors of sexual interest in adults with a medium The use of pupil dilation as a measure of sexual interest has and a large expected effect to equal 118 and 57 participants, a long history in the psychophysiology literature but has only respectively. Recruitment materials called for participants for recently achieved a level of consistency in findings that sup- a study ‘comparing eye tracking with other computer-based ports its validity. Using a somewhat rudimentary measurement measures of sexual interests.’ Participants were recruited using approach, Hess, Seltzer, and Shlien (1965) found evidence of a nontargeted approach via the university’s online recruitment pupil dilation to orientation-consistent stimuli among a small system, in addition to more purposive sampling through the uni- sample of gay and straight men. Other researchers (e.g., Scott, versity’s student LGBT? society social media page to increase Wells, & Wood, 1967), however, failed to replicate their gen- numbers of gay and bisexual men in the samples (for a similar eral findings. As a result, this method has been relatively under- approach, see Rieger & Savin-Williams, 2012). researched until recently, when Rieger and Savin-Williams (2012) found, using a large sample, that men’s pupils dilated to erotic video stimuli in line with their sexual interests. Rieger et al. (2015) further established the construct validity of pupil Attard-Johnson et al. (2017) presented a detailed examination of the dilation as a measure of sexual interest by demonstrating a cor- pupil dilation data of 100 participants from this data set. 123 640 Arch Sex Behav (2018) 47:637–649 Mean age was 22.2 years (SD = 5.7; range, 18–50 years). All Measures participants had completed secondary education. Ethnicity was not recorded in this study, but the sample was recruited through Pupil Dilation and Viewing Time Tasks advertising in a majority White but ethnically diverse university Participants completed four tasks (one free-viewing and three campus. Participants reported the degree to which they were attracted to male and female adults. The majority were exclu- viewing time tasks) while their pupil responses were recor- ded. Before commencing each task, the system was calibrated sively attracted to female adults (59.8%), 13.7% were predom- inantly attracted to females, 4.9% were attracted equally to using the standard nine-point fixation Eyelink procedure. male and female adults, 2.0% were predominantly attracted to male adults, and 19.6% were exclusively attracted to male adults. Free Viewing of Beach Set Stimuli In this task, participants All participants had normal or corrected-to-normal vision. were presented with the natural beach scenes. A free-viewing paradigm was adopted whereby participants were instructed to view the images as ‘naturally as they normally would’ (for Stimuli similar approaches, see Attard-Johnson et al., 2016; Hall, Hogue, &Guo, 2011). Each trial began with a fixation dot, which allowed Pictorial stimuli used across measures were from three sources. for drift correction, and ensured that participants attended to the TheNotRealPeople(NRP)set (clothed version; Laws & Gress, center of the display when the trial began. The trial began with a 2004) are composite images of males and females previously gray screen which was displayed for 1 s, followed by the stimulus developed for use in research involving sexual offenders. Images displayed for 10 s, followed by another gray screen for 1 s. Each in the NRP set are presented on a neutral background, classi- participant viewed all 25 images once with order randomized fied according to their secondary sexual characteristics (Tan- across participants. ner, 1973), and represent all five Tanner stages. A second set of images (Morph set;O Ciardha, 2010) provided a greater num- Viewing Time of Beach Set Stimuli The eye movement ber of adult (Tanner 5) stimuli than were available using the setup and procedure was identical to the previous task, except NRP set alone. These images are realistic but do not represent for the following. Scenes without person content were excluded real people, as faces are morphed composites of different faces. from the beach set stimuli. Instead of being instructed to view This set included control images of big cats. The third set of images ‘naturally as they normally would,’ participants were images (Beach set; Attard-Johnson et al., 2016) consisted of instructed to rate the sexual appeal of these targets on a seven- stock photographs portraying adult and prepubescent (Tanner point scale, using the number keys on a standard keyboard. 1) males and females in natural beach scenes (5 images in each Once a response was registered, the next trial was presented. of these four categories) and a set of control landscapes without any person content (5 scenes). People were portrayed in swim Viewing Time of NRP Set Stimuli This task employed the or leisure wear. same method and procedure as the previous viewing time task, except that the scene stimuli were replaced with images from the NRP picture set. A total of 40 images were used (four males Apparatus and four females at each Tanner stage). Pupil Dilation Tasks that included measurement of pupil Viewing Time of Morph Set Stimuli The procedure was dilation were displayed using SR-Research ExperimentBuilder identical to the previous viewing time tasks, except that a total software (version 1.1.0) on a 21’’ color monitor, with a screen of 36 computer-modified morph set stimuli were used (18 adult resolution of 10249 768 pixels. Eye movements were recorded males and 18 adult females). using an SR-Research Eyelink 1000 eye-tracking system. The Eyelink 1000 was running at a 1000-Hz sampling rate, a spatial Choice Reaction Time Task (CRT) resolution of\.01 of visual angle, a gaze position accuracy of\ .5, and a pupil size resolution of .1% of diameter. The Eyelink The CRT was presented using SuperLab4 stimulus presenta- 1000 system works by measuring corneal reflection and dark tion software and asked participants to identify as quickly as pupil with a video-based infrared camera. This system computes possible the location of a dot that appeared in one of five posi- the number of camera pixels that are occluded by partici- tions on an image of a person (or big cats in the practice con- pants’ pupils and records the area of the pupil as an integer that dition). Stimuli for this task, and for all other reaction time ranges from 400 to 16,000 units. During the recording of pupil tasks, were presented on a Dell E6520 laptop with a 15.6’’ color size, participants placed their heads on a chinrest approxi- monitor and a screen resolution of 13669 768 pixels. Responses mately 60 cm from the display monitor. were via a Cedrus RB-530 response pad. A white dot (approx- 123 Arch Sex Behav (2018) 47:637–649 641 imately 2.3 mm in circumference) was superimposed on each cross presented centrally for 500 ms. Participants could not con- of five positions on the images, yielding five versions of each tinue to the next trial until they had answered correctly. Stimuli image, where the white dot was located either in the top right, were presented at least twice within each block. top left, bottom right, bottom left, or middle of the image. Each The first f-IAT block presented stimuli, one at a time, from stimulus presentation was preceded by a centrally located fixa- the concept categories (adult and child images) and partici- tion cross for 500 ms. Stimuli remained on the screen until par- pants were instructed to identify them using the left and right ticipants responded. Participants received no feedback on errors, buttons on the response pad. Correct category headings were except for the first 20 practice trials. presented in the corresponding left and right corner of the Stimuli were taken from the NRP set and the morph set. screen. In the second block, participants categorized word Images were classified by age as pre-/early pubertal (Tanner 1 stimuli, again using the left and right buttons, as belonging to and 2), late pubertal (Tanner 3 and 4), and adult (Tanner 5), and one of the attribute categories. Sexual words were‘‘lust,’’‘‘lick,’’ by sex (female and male) to yield six trial types. Trial types were ‘‘kiss,’’‘‘naked,’’‘‘orgasm,’’‘‘arouse,’’and‘‘attractive,’’whilenon- presented together in small blocks or clusters of ten images. sexual words were ‘‘ugly,’’ ‘‘cold,’’ ‘‘dull,’’ ‘‘avoid,’’ ‘‘bland,’’ There were four clusters of each trial type, yielding 40 images ‘‘boring,’’and‘‘unattractive.’’In the third block, the left-hand con- of each trial type. Trials were randomized in each cluster, and cept category from block one was paired with the left-hand cluster order was randomized with the constraint that one of attribute category from block two (e.g., male and sexual), and each trial type cluster had to be presented before seeing a second likewise for the right-hand concept and attribute categories (e.g., of any type and so on. Response times and response correctness female and nonsexual). Only adult images were presented. The were recorded. fourth block was identical to the third except that only child images were presented. Pictorial Modified Stroop Task (P-MST) In the fifth block, the order of the concept pairs was swapped, whereby female might now appear on the left and male on the The P-MST asked participants to identify as quickly as possible, right. Both adult and child images were included in this block. by pressing colored buttons on the response pad, the color in In blocks six and seven, participants categorized both images which an image was presented, again using SuperLab4. Images and words by concept or attribute but with this new arrange- from the NRP and the morph sets had been tinted red, green, ment of the concept pairs. In block six, images were of adults, blue, and yellow. Four colored versions of twelve images each and in block seven images were of children. were used to create blocks of child male, child female, adult Participants’ response times to blocks 3/4 and 6/7 were ana- male, and adult female images, resulting in 48 images in each lyzed as differences between them should indicate differences block. Unlike the CRT, blocks of child stimuli combined images in the relative strength of associations between the concepts of male and female and the attributes of sexual and nonsexual. rated as belonging to Tanner stages 1–4. At the beginning of the task, participants completed a block of 16 practice trials fea- Furthermore, the strength of these associations could be exam- turing pictures of big cats. Feedback was given for incorrect ined to explore the impact of child stimuli. Order was random- responses for practice trials but not for the experimental trials. ized for participants so that some were presented with male Presentation of stimuli was preceded by a centrally located fix- sexual associations first, whereas others were presented first ation cross for 500 ms. Stimuli remained on the screen until with female-sexual pairings. Blocks three and six contained participants responded. A blocked design was adopted, whereby adult stimuli for all participants, with child stimuli in blocks all adult male images were presented together, as were all adult four and seven. females, and so on. Order of block presentation was randomized. As with the CRT, response times and response correctness were Interest in Child Molestation (ICM) Scale and Problematic recorded. Sexual Behavior or Fantasy Factorial Implicit Association Test (f-IAT) The scenarios used in this study were identical to the original ICM scale developed by Gannon and O’Connor (2011), with minor The f-IAT examined the strength of association between male/ changes to question order and wording. The scale presented five female concept categories and sexual/nonsexual attribute cate- scenarios describing sexual abuse with a child (age and gender gories. Stimuli for the concept category were images from the unspecified). Participants were asked to imagine themselves in NRP and morph sets. The f-IAT was based on an IAT used by the situation described. They were then asked to answer the O Ciardha and Gormley (2013), but included images of adults following questions on a seven-point scale: (1)‘‘In this situation, and children resulting in an extra‘‘factorial’’element. Twelve how sexually aroused would you be?’’(anchor points on the scale male and 12 female images were used, half of adults (Tanner 5) were ‘‘not at all sexually aroused’’ and ‘‘very strongly sexually and half of children (Tanner 1–3). The f-IAT was administered aroused’’); (2) ‘‘In this situation, how much would you enjoy over seven blocks. Each stimulus was preceded by a fixation getting your way?’’ (anchors of ‘‘would not enjoy it at all’’ and 123 642 Arch Sex Behav (2018) 47:637–649 ‘‘would greatly enjoy it’’; and (3) ‘‘If you found yourself in a 25th and 75th percentiles within that experimental condition similar situation, would you have done the same?’’ (anchors of for that individual. All reaction time data were converted to ‘‘would definitely not have done the same’’and‘‘would definitely ipsative z-scores. Ipsative z-scores present an individual’s have done the same’’). The total possible score on the ICM scale mean response times for a given condition in terms of numbers ranged from 15 to 105, with the minimum score indicating an of standard deviations (based on their own data) from their over- emphatic rejection of arousal, enjoyment, or propensity across all all mean response time. Therefore, if their response times on a child molestation conditions. P-MST to images of adult females yields an ipsative z-score Additional questions were included to examine prior sex- of ? 2, they have been responding slower to images of adult ual offending behavior and sexual fantasy involving children females than their overall mean reaction time by a magni- among participants. They were asked if they (1) had ever know- tude of two standard deviations. ingly and deliberately viewed pornographic material contain- Calculating ipsative z-scores for the f-IAT differs from the ing individuals below the age of consent, (2) had, since the age more standardD algorithm approach, suggested by Greenwald, of 18, found the thought of sex or sexual contact with a person Nosek, and Banaji (2003). However, this alternative has been 15 years of age or younger exciting or arousing, (3) had sexual used previously with a similar IAT (O Ciardha & Gormley, contact with a child under 13 who was more than 5 years younger 2013) and has the strength of being directly comparable with than them since the age of 16, or (4) had sexual contact with an the approach taken with the P-MST and CRT tasks. Addition- individual aged 13–15 who was more than 5 years younger since ally, the ipsative z-score approach allowed the separate exam- the age of 18. Only the results of question 1 are presented in the ination of the blocks in which adult and child stimuli were current study. presented. Procedure Viewing Time Tasks The study was reviewed and approved by the School of Psychol- As with the reaction time data, viewing time task response times ogy Research Ethics Committee (REF 20122341). Participants were converted to ipsative z-scores. No outlier removal was car- were seated in a windowless room with constant artificial light- ried out as there were fewer experimental trials in each condition ing. Participants completed tasks involving pupil dilation first compared with the reaction time tasks, and participants were not and in the following order: free viewing of beach set, viewing instructed to respond rapidly, meaning even conservative outlier time of beach set, viewing time of NRP set, and viewing time of removal risks removing trials that are not inconsistently slow in morph set. Participants then completed the reaction time tasks the context of the task instructions. (CRT, P-MST, and IAT). The order of reaction time tasks was counterbalanced. Finally, participants completed a brief ques- Pupil Dilation tionnaire to assess sexual interests followed by the ICM scale and questions about problematic sexual behavior or fantasy. In Observers’ pupillary responses to each stimulus category were a small number of cases, participants did not complete partic- calculated as a percentage change from observers’ overall pupil ular tasks or data did not record properly for technical reasons. mean. For this, pupillary responses were first computed by tak- We report sample sizes for individual analyses to reflect this. ing the mean pupil area at each fixation, averaged across the The average length of the entire procedure including obtaining duration of a stimulus display. An overall mean, across all stim- consent, questionnaires, completion of indirect tasks, calibration uli in all conditions, was then computed from these values for and administration of eye tracking/pupil dilation paradigms, and each participant. The percentage difference (i.e., an increase or debriefing was approximately 90 min per participant. decrease) in pupil area for each stimulus category (e.g., adult males) from the overall mean was then computed, using the Data Preparation formula: (mean pupil area for category 9 100)/overall pupil mean. Accordingly, a score of 100% indicates that the pupillary Reaction Time Tasks response to a stimulus category does not differ from the overall mean. Scores above or below this value indicate comparatively Reaction time data were calculated based on correct responses larger or smaller pupil sizes (for similar approaches, see Attard- only and were trimmed of outliers by removing times more Johnson et al., 2016; Laeng & Falkenberg, 2007). extreme than three times the interquartile range beyond the 123 Arch Sex Behav (2018) 47:637–649 643 carry out preliminary investigations to examine whether par- Results ticipants who appeared to demonstrate sexual interest in younger Assessing Sexual Interest in Adults individuals—through increased attention, stronger sex associa- tions, or greater pupil dilation—also self-reported greater interest For each of the measures (reaction times, viewing times, and in child molestation or were more likely to report having previ- ously used pornographic material containing children. pupil dilation), an index of preferred adult sex was calculated by subtracting the mean responses to adult female stimuli from We coded our predictor variables into binary values indicat- ing whether individuals’ strongest average response was to adult the mean responses to adult male stimuli (see Table 1). Each of these ten indices were positively correlated, with effect sizes stimuli (Tanner 5) or to younger stimuli. Table 4 shows the percentage of participants with a stronger response to younger ranging from small to large (Cohen, 1988; see Table 2). Sexual stimuli across all tasks alongside rank-biserial correlations with interest measured as a five-point continuous variable (exclu- the ICM. The binary predictors of adult or child interest were sively straight, given a value of 1, to exclusively gay, given a entered into a multiple linear regression predicting ICM scores value of 5) positively correlated with the index of preferred adult gender for each measure (rs .39–77; see Table 2). (see Table 5). Overall, the model accounted for 21% of the vari- ance in ICM scores. This reflects a moderate effect size. A multiple linear regression model incorporating all indices of preferred adult gender explained 75% of the variance in sex- In our sample, there were too few participants who admit- ted having viewed pornographic material containing children ual interest (see Table 3). This amount of variance explained equates to a large effect size. However, it also demonstrates to examine multiple predictors of that outcome in a logistic regression. Therefore, we examined the relationship between that there was still some error in the ability of the indices of indirect measures and use of pornographic material contain- preferred gender to predict self-reported sexual interest. We ing children using a t test. The dependent variable in this case explored whether a source of error might stem from cases was the total number of tasks showing a stronger response to where individuals reported non-exclusive sexual interests, child stimuli compared to adults. Across the eight paradigms, by examining whether exclusivity moderated the relation- individuals who admitted use of pornographic material con- ship between predicted values in the regression model and observed sexual interest (all variables standardized), using taining children had a greater number of strongest responses to child stimuli (M = 3.8, SD = 1.5) compared with individuals the PROCESS macro for SPSS (model 1; Hayes, 2013). The analysis (n = 96) showed that the relationship between pre- whoreported nouse (M= 2.8, SD= 1.2). This difference,- .99, BCa 95% CI [- 1.91, - .06], represented a large effect size, dicted and observed values was weaker for those with non- exclusive sexual interests, b = .32; SE = .18; BCa 95% CI albeit with a wide confidence interval; t(94)=- 2.57, d = .80 (effect size calculated using supplementary materials from [- .04, .67], compared with exclusive sexual interests, b = .91; SE = .05; BCa 95% CI [.80, 1.01]. Lakens, 2013). Assessing Sexual Interest in Children Discussion The mean interest in child molestation (ICM) score for this sample was 20.6 (SD = 9.3; range 15–59), with just over half In our sample, indices of sex preference to adult stimuli across tasks were consistently related to one another and to self-reported the sample (51%) emphatically rejecting an interest in child sexual abuse (i.e., scoring 15). Additionally, 11.8% (n = 13) sexual interest toward adults, consistent with our hypotheses. All correlations between indices were positive though effect sizes admitted knowingly and deliberately viewing pornographic material containing individuals below the age of consent. varied from small to large. Correlations between indices of sex preference and self-reported sexual interest toward adults were Despite non-minimal responding in the ICM and admission again positive, with effect sizes ranging from moderate to of the use of pornographic material containing children by large. These relationships demonstrated evidence of the con- some participants, it is unlikely that the sample would contain vergent and concurrent validity of latency-based tasks and of enough individuals with a clear sexual interest in children to robustly examine whether the tasks were valid measures of pre- pupil dilation. ferred age as well as preferred sex. However, it was possible to 2 3 For reasons of parsimony, we used the word preferred with reference Some previous studies combine Tanner 4 and 5 images as representing to gender or age in describing whether response time or pupil dilation mature (e.g., Ebsworth & Lalumiere, 2012) or have omitted Tanner 4 tasks indicated greater sexual interest to stimuli of a particular sex or age images entirely (e.g., Banse, Schmidt, & Clarbour, 2010). We were unable across individuals. We used the word to indicate the relative strength of to isolate Tanner 4 images across all paradigms presented. We therefore associations, pupil dilations, etc., to different stimuli, not to imply volun- examined only whether individuals responded most strongly to Tanner 5 tary choice in sexual orientation (see Committee on Lesbian and Gay versus all other Tanner stages. We examine implications in the Discussion Concerns, American Psychological Association, 1991). section. 123 644 Arch Sex Behav (2018) 47:637–649 Table 1 Samples sizes, grand means, standard deviations, and ranges for indices of sexual interest in adults nM (SD) Range Min Max Reaction time measures P-MST 98 - .12 (.63) - 1.76 1.68 CRT 100 - .17 (.80) - 3.33 2.57 f-IAT 100 - .32 (.67) - 1.63 1.31 Viewing time tasks NRP 102 - .81 (1.34) - 3.21 2.86 Morph 100 - .41 (.96) - 1.88 1.64 Beach 100 - .39 (.93) - 2.21 1.54 Pupil dilation NRP 100 .27 (6.65) - 15.93 17.85 Morph 100 - 1.68 (5.18) - 18.14 8.46 Beach (VT) 100 - 2.6 (7.14) - 18.00 14.77 Beach (free) 100 - 3.51 (6.48) - 19.13 14.69 P-MST, pictorial modified Stroop task; f-IAT, factorial Implicit Association Test; NRP, not real people image set (Laws & Gress, 2004); VT, viewing time; reaction and viewing time values represent the mean difference in ipsative z-scores between responses to adult male versus adult female stimuli; pupil dilation values represent the mean difference between pupillary responses to adult male versus adult female stimuli as a percentage of individual mean pupil size. For all measures, negative means suggest greater sexual interest in female adults Combined indices of sex preference to adult stimuli predicted Taken together, and interpreted in the context of a low base rate 75% of the variance in self-reported sexual interest in adults. of pedophilia and hebephilia in community samples, these While some predictors did not uniquely contribute to the results are very preliminary indicators of concurrent validity model’s overall ability to predict self-reported sexual interest, of the tasks as measures of sexual interest in children. those that did spanned reaction time, viewing time, and pupil Our findings offer evidence that indirect measures of sexual dilation approaches. In a post hoc analysis, we found that exclu- interest operate predictably and in a manner that is consistent sivity of sexual interest moderated the strength of the relation- with one another, particularly with adult stimuli. This was true ship between the predicted values and the observed value in our of the cognitive and psychophysiological paradigms exam- regression model above. This finding appears consistent with ined in the current study, at least in the conditions experienced the finding by Rieger et al. (2015) of weaker correspondence by our participants. Different processes relating to sexual atten- between indices of arousal in bisexual men compared to gay tion or arousal are likely to underpin these effects depending on or straight men. the task (O Ciardha, 2011). Pupil dilation, for example, is likely Our examination of measures as predictors of sexual inter- to reflect increased activity in the sympathetic nervous system est in children was exploratory given the expectation of low (Bradley, Miccoli, Escrig, & Lang, 2008). IAT, on the other numbers of participants with sexual interest in children. To hand, measures the strength of association between simple con- determine whether indirect task performance suggested a cepts held in memory and therefore possibly reflects schematic sexual interest in children, we created post hoc dichotomous relationships in ‘implicit sexual memory’ (for a discussion of variables indicating whether the strongest response on any explicit and implicit sexual memory, see Spiering & Everaerd, given indirect measure was to child or adult stimuli. Table 5 2007). shows the percentage of individuals showing a strongest response The CRT and P-MST appear to be measures of selective to child stimuli across indirect measures, and how measures cor- attention. However, both these tasks in the current study did relate with ICM scores. It is noteworthy that the percentage not randomly present individual trials but rather combined of participants showing a stronger response to children varied stimuli of the same trial type into blocks or clusters of stimuli. considerably depending on the task. This is a point we return to This follows recommendations for maximizing modified Stroop later. Indirect tasks performance, dichotomized in this way effects with appetitive stimuli (specifically addiction Stroop explained 21% of the variance in ICM scores, corresponding to tasks, see Cox, Fadardi, & Pothos, 2006). With a blocked design, a moderate effect size. Knowingly and deliberately viewing it is not possible to examine whether differences in latencies are due to an instantaneous capturing of attention or because of pornographic material containing children was related to hav- ing stronger response to children in a greater number of tasks. carryover effects once a given stimulus has been erased. As a 123 Arch Sex Behav (2018) 47:637–649 645 Table 2 Correlation coefficients between indices of preferred adult sex and self-reported adult sexual interests 1 2345 678910 Reaction time measures 1. P-MST 2. CRT r .62 n 98 BCa 95% CI [.44, .74] 3. f-IAT r .49 .52 n 98 100 BCa 95% CI [.33, .61] [.40, .64] Viewing time tasks 4. NRP r .49 .43 .53 n 98 100 100 BCa 95% CI [.38, .59] [.29, .56] [.38, .66] 5. Morph r .55 .40 .50 .83 n 96 98 98 100 BCa 95% CI [.41, .66] [.26, .53] [.31, .65] [.76, .89] 6. Beach r .43 .37 .40 .67 .72 n 96 98 98 100 100 BCa 95% CI [.20, .61] [.11, .58] [.21, .57] [.55, .77] [.61, .81] Pupil dilation 7. NRP r .42 .37 .40 .64 .65 .48 n 96 98 98 100 100 100 BCa 95% CI [.28, .53] [.22, .53] [.23, .54] [.49, .75] [.52, .76] [.32, .62] 8. Morph r .44 .46 .53 .68 .68 .55 .66 n 96 98 98 100 100 100 100 BCa 95% CI [.31, .57] [.33, .60] [.39, .65] [.57, .77] [.57, .77] [.41, .68] [.50, .77] 9. Beach (VT) r .14 .29 .22 .39 .43 .45 .32 .49 n 96 98 98 100 100 100 100 100 BCa 95% CI [- .04, .32] [.04, .52] [.03, .40] [.23, .53] [.28, .57] [.30, .59] [.13, .49] [.34, .62] 10. Beach (free) r .15 .20 .23 .30 .39 .31 .29 .39 .38 n 96 98 98 100 100 100 100 100 100 BCa 95% CI [- .002, .31] [.02, .39] [.05, .40] [.11, .48] [.24, .53] [.13, .46] [.11, .46] [.23, .54] [.17, .54] Sexual orientation r .59 .46 .65 .72 .77 .58 .58 .69 .39 .44 n 98 100 100 102 100 100 100 100 100 100 BCa 95% CI [.46, .69] [.32, .58] [.52, .75] [.58, .83] [.67, .85] [.44, .69] [.44, .70] [.59, .77] [.24, .52] [.29, .59] P-MST, pictorial modified Stroop task; f-IAT, factorial Implicit Association Test; NRP, not real people image set (Laws & Gress, 2004); VT, viewing time; BCa 95% CI, 95% confidence intervals based on 5000 resamples using the bias-corrected and accelerated (BCa) bootstrap method result, while the current study provided evidence of criterion among our sample (83%) was the pupil dilation method using validity across this battery of tasks by demonstrating their con- NRP stimuli. The tasks using NRP stimuli were the only tasks vergence, research examining their discriminant validity is which generated response data for each Tanner stage individ- needed to better understand the processes underpinning indi- ually. Inspection of the raw data showed that the percentage vidual tasks. was driven in large part by the number of participants (31%) In the current study, tasks containing child images did not showing the greatest average pupil dilation to Tanner 4 stimuli. all have identical groupings of stimuli by stage of sexual devel- Ebsworth and Lalumie `re (2012) described Tanner 4 as repre- opment. This was a consequence of selecting tasks that had senting‘‘individuals who are on the cusp of reaching full sexual previously been piloted or otherwise tested within our labo- maturity’’ (p. 164). As a result, a sexual interest in Tanner 4 ratory as part of existing research streams. It is noteworthy that images may not be a particularly useful index of forensically the measure suggesting the highest rates of interest in children relevant sexual interest, especially given the low average age of 123 646 Arch Sex Behav (2018) 47:637–649 Table 3 Summary of multiple linear regression analysis examining Table 5 Summary of multiple linear regression analysis examining whether indices of preferred adult sex predict sexual orientation (on a whether indices of preferred age predict ICM scores five-point scale) B SE B b BCa 95% CI B SE B b BCa 95% CI Reaction time measures Reaction time measures P-MST 1.54 2.40 .08 [- 3.06, 6.74] P-MST .59 .22 .23 [.15, 1.03] CRT 1.79 1.83 .10 [- 1.71, 5.45] CRT - .22 .16 - .10 [- .54, .09] f-IAT 1.20 1.87 .06 [- 2.54, 4.88] f-IAT .55 .14 .23 [.25, .83] Viewing time tasks Viewing time tasks NRP - .97 1.86 - .05 [- 4.83, 2.67] NRP .22 .16 .19 [- .11, .56] Beach 10.32 5.46 .31 [.13, 22.00] Morph .44 .22 .27 [.001, .87] Pupil dilation Beach - .11 .13 - .07 [- .36, .12] NRP 5.80 1.37 .23 [3.53, 8.36] Pupil dilation Beach (VT) 2.28 2.66 .08 [- 2.10, 6.90] NRP .01 .02 .02 [- .04, .05] Beach (free) - 1.02 1.99 - .04 [- 5.23, 3.09] Morph .05 .03 .18 [- .01, .12] R .21 Beach (VT) - .002 .02 - .01 [- .04, .03] F 2.94 Beach (free) .03 .02 .14 [.002, .07] n 96 R .75 P-MST, pictorial modified Stroop task; f-IAT, factorial Implicit Asso- F 25.82 ciation Test; NRP, not real people image set (Laws & Gress, 2004); VT, n 96 viewing time; B, unstandardized regression coefficient; b, standardized regression coefficient; BCa 95% CI, 95% confidence intervals of B based P-MST, pictorial modified Stroop task; f-IAT, factorial Implicit Asso- on 5000 resamples using the bias-corrected and accelerated (BCa) ciation Test; NRP, not real people image set (Laws & Gress, 2004); VT, bootstrap method viewing time; B, unstandardized regression coefficient; b, standardized regression coefficient; BCa 95% CI, 95% confidence intervals of B based on 5000 resamples using the bias-corrected and accelerated (BCa) our participants. Despite this limitation, however, the pupil dila- bootstrap method tion method using NRP stimuli was a predictor of ICM scores, with inspection of descriptive data indicating that participants showing the strongest pupil dilation to Tanner 5 stimuli also had the lowest ICM scores. Future studies using indirect Table 4 Percentage of participants with a stronger average indirect task response to child stimuli and rank-biserial correlation between having a measures should be designed, where possible, to measure the stronger response to children and ICM scores relative strength of response to stimuli across the full range of n Percentage of Correlation with ICM sexual maturity. strongest response The most robust findings of our study were those estab- r BCa 95% CI rb to child stimuli (%) lishing convergent and concurrent validity across tasks mea- suring sexual interest using adult stimuli, particularly where Reaction time measures participants reported exclusive sexual interests in one sex or P-MST 98 29.6 .07 [- .14, .28] another. The more exploratory findings examining sexual CRT 100 56.0 .08 [- .12, .26] interest in children will require corroboration with future stud- f-IAT 100 44.0 .04 [- .16, .24] ies containing greater number of pedophilic and hebephilic par- Viewing time tasks ticipants. However, establishing criterion validity of sexual inter- NRP 102 40.2 - .02 [- .21, .18] est tasks for forensic use based on age-appropriate sexual interest Beach 100 8.0 .25 [.003, .44] using non-offending samples is an important step. Seto (2017), Pupil dilation for example, outlined a multidimensional view of sexual orien- NRP 100 83.0 .32 [.16, .45] tation, defining sexual orientation as a ‘‘stable tendency to pref- Beach (VT) 100 13.0 .10 [- .11, .30] erentially orient—in terms of attention, interest, attraction, and Beach (free) 100 14.0 .05 [- .15, .26] genital arousal—to particular classes of sexual stimuli’’ (p. 3). P-MST, pictorial modified Stroop task; f-IAT, factorial Implicit Asso- Seto (2017) argued that while sex is the usual dimension ciation Test; NRP, not real people image set (Laws & Gress, 2004); VT, along which sexual orientation is discussed, age, or more specif- viewing time; r , rank-biserial correlation coefficient; BCa 95% CI, rb ically physical and sexual maturity, represents a dimension on 95% confidence intervals based on 5000 resamples using the bias-cor- rected and accelerated (BCa) bootstrap method which people may hold a stable orientation. He referred to 123 Arch Sex Behav (2018) 47:637–649 647 Attard-Johnson, J., Bindemann, M., & O Ciardha, C. (2016). Pupillary variations in individual orientation on this age dimension as response as an age-specific measure of sexual interest. Archives of chronophilias, which include age preferences to prepubescent Sexual Behavior, 45(4), 855–870. https://doi.org/10.1007/s10508- children (pedophilia), pubescent children (hebephilia), young 015-0681-3. adults (teleiophilia), etc. Within this multidimensional model, Attard-Johnson, J., Bindemann, M., & O Ciardha, C. (2017). Heterosexual, homosexual, and bisexual men’s pupillary responses to persons at sexual interests may differ across individuals in terms of what is different stages of sexual development. Journal of Sex Research, attractive, but may not differ in how that attraction impacts atten- 54(9), 1085–1096. https://doi.org/10.1080/00224499.2016.1241857. tion and arousal (see also O Ciardha, 2011). Therefore, results Babchishin, K. M., Nunes, K. L., & Hermann, C. A. (2013). The validity when measuring teleiophilic sexual interest among men (e.g., of Implicit Association Test (IAT) measures of sexual attraction to children: A meta-analysis. Archives of Sexual Behavior, 42(3), 487– task validity) should generalize to the testing of male sexual inter- 499. https://doi.org/10.1007/s10508-012-0022-8. est toward other age groups. If they do not, there are implications Banse, R., Schmidt, A. F., & Clarbour, J. (2010). Indirect measures of for the conceptualization of pedophilia, for example, as dimen- sexual interest in child sex offenders: A multimethod approach.Crim- sionally related to a sexual interest in adults. inal Justice and Behavior, 37(3), 319–335. https://doi.org/10.1177/ Data were collected in the current study under non-adver- Barnes-Holmes, D., Barnes-Holmes, Y., Power, P., Hayden, E., Milne, R., sarial conditions. As a result, we cannot be certain which mea- & Stewart, I. (2006). Do you really know what you believe? Devel- sures are most resilient to faking or whether there are certain oping the Implicit Relational Assessment Procedure (IRAP) as a direct conditions under which some or all measures do not correspond measure of implicit beliefs. The Irish Psychologist, 32(7), 169–177. Bradley, M. M., Miccoli, L., Escrig, M. A., & Lang, P. J. (2008). The pupil as a to an individual’s sexual interest (i.e., discriminant validity). measure of emotional arousal and autonomic activation. Psychophys- However, we are least enthusiastic about the potential utility of iology, 45(4), 602–607. https://doi.org/10.1111/j.1469-8986.2008.006 the viewing time approach, as it is the most transparent method- 54.x. ology. Schmidt et al. (2017) state similar concerns and lament Cohen, J. (1988). Statistical power analysis for the behavioral sciences. New York, NY: Routledge Academic. the lack of fakeability studies. While knowledge of reaction Committee on Lesbian and Gay Concerns, American Psychological time taskssuchasthe P-MST, CRT, and f-IAT would theoret- Association. (1991). Avoiding heterosexual bias in language. Amer- ically also allow people to manipulate their responses, the fact ican Psychologist, 46(9), 973–974. https://doi.org/10.1037/0003-06 that rapid responses are required makes it less likely that par- 6X.46.9.973. Cox, W. M., Fadardi, J. S., & Pothos, E. M. (2006). The addiction-Stroop ticipants could produce a coherent but fake pattern of respond- test: Theoretical considerations and procedural recommendations. ing. Given that pupil dilation is an automatic response to activity Psychological Bulletin, 132(3), 443–476. https://doi.org/10.1037/ in the sympathetic nervous system (Bradley et al., 2008), it is 0033-2909.132.3.443. likely to be the task that is least susceptible to deliberate manipu- Cvencek, D., Greenwald, A. G., Brown, A. S., Gray, N. S., & Snowden, R. J. (2010). Faking of the Implicit Association Test is statistically detec- lation (Laeng et al., 2012). Future research should address issues table and partly correctable. Basic and Applied Social Psychology, of fakeability, discriminant validity, and of course include 32(4), 302–314. https://doi.org/10.1080/01973533.2010.519236. participants with an admitted sexual interest in children. How- Dombert, B., Antfolk, J., Kallvik, L., Zappala `, A., Osterheider, M., Mokros, ever, taken together our results are the first to demonstrate strong A., et al. (2015). Identifying pedophilic interest in sex offenders against children with the Indirect Choice Reaction Time Task. European Jour- evidence of the convergent and concurrent validity of viewing nal of Psychological Assessment. https://doi.org/10.1027/1015-5759/ time, reaction time, and pupillometric approaches to the mea- a000293. surement of sexual interest. Ebsworth, M., & Lalumie `re, M. L. (2012). Viewing time as a measure of bisexual sexual interest. Archives of Sexual Behavior, 41(1), 161– Acknowledgements This work was supported by a research grant from 172. https://doi.org/10.1007/s10508-012-9923-9. the National Organisation for the Treatment of Sexual Abusers (NOTA) Gannon, T. A., & O’Connor, A. (2011). The development of the Interest in to Caoilte O Ciardha and Markus Bindemann. Child Molestation Scale. Sexual Abuse: A Journal of Research and Treatment, 23(4), 474–493. https://doi.org/10.1177/1079063211412 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons. Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding org/licenses/by/4.0/), which permits unrestricted use, distribution, and repro- and using the Implicit Association Test: I. An improved scoring duction in any medium, provided you give appropriate credit to the original algorithm. Journal of Personality and Social Psychology, 85(2), author(s) and the source, provide a link to the Creative Commons license, and 197–216. https://doi.org/10.1037/0022-3514.85.2.197. indicate if changes were made. Gress, C. L. Z., Anderson, J. O., & Laws, D. R. (2011). Delays in attentional processing when viewing sexual imagery: The development and com- parison of two measures. Legal and Criminological Psychology, 18(1), 66–82. https://doi.org/10.1111/j.2044-8333.2011.02032.x. Hall, C., Hogue, T., & Guo, K. (2011). Differential gaze behavior towards References sexually preferred and non-preferred human figures. Journal of Sex Research, 48(5), 461–469. https://doi.org/10.1080/00224499.2010. Abel, G. G., Huffman, J., Warberg, B., & Holland, C. L. (1998). Visual reaction time and plethysmography as measures of sexual interest in Hanson, R. K., & Morton-Bourgon, K. (2004). Predictors of sexual recidi- child molesters. Sexual Abuse: A Journal of Research and Treat- vismanupdatedmeta-analysis. Ottawa, ON: Public Works and Govern- ment, 10(2), 81–95. https://doi.org/10.1023/A:1022063214826. ment Services Canada. 123 648 Arch Sex Behav (2018) 47:637–649 Hayes, A. F. (2013). Introduction to mediation, moderation, and con- roImage, 43(2), 337–345. https://doi.org/10.1016/j.neuroimage.20 ditional process analysis: A regression-based approach. New York, 08.07.059. NY: Guilford. Price, S. A., & Hanson, R. K. (2007). A modified Stroop task with sexual Hess, E. H., Seltzer, A. L., & Shlien, J. M. (1965). Pupil response of offenders: Replication of a study. Journal of Sexual Aggression, hetero- and homosexual males to pictures of men and women: A 13(3), 203–216. https://doi.org/10.1080/13552600701785505. pilot study. Journal of Abnormal Psychology, 70(3), 165–168. https:// Rieger, G., Cash, B. M., Merrill, S. M., Jones-Rounds, J., Dharmavaram, doi.org/10.1037/h0021978. S. M., et al. (2015). Sexual arousal: The correspondence of eyes and Imhoff, R., Schmidt, A. F., Nordsiek, U., Luzar, C., Young, A. W., & Banse, genitals. Biological Psychology, 104, 56–64. https://doi.org/10.1016/ R. (2010). Viewing time effects revisited: Prolonged response latencies j.biopsycho.2014.11.009. for sexually attractive targets under restricted task conditions. Archives Rieger, G., & Savin-Williams, R. C. (2012). The eyes have it: Sex and of Sexual Behavior, 39(6), 1275–1288. https://doi.org/10.1007/s10508- sexual orientation differences in pupil dilation patterns. PLoS ONE, 009-9595-2. 7(8), e40256. https://doi.org/10.1371/journal.pone.0040256. Kalmus, E., & Beech, A. R. (2005). Forensic assessment of sexual interest: A Ro ¨ nspies, J., Schmidt, A. F., Melnikova, A., Krumova, R., Zolfagari, A., review. Aggression and Violent Behavior, 10(2), 193–217. https://doi. & Banse, R. (2015). Indirect measurement of sexual orientation: org/10.1016/j.avb.2003.12.002. Comparison of the implicit relational assessment procedure, viewing Laeng, B., & Falkenberg, L. (2007). Women’s pupillary responses to sexually time, and choice reaction time tasks. Archives of Sexual Behavior, significant others during the hormonal cycle. Hormones and Behavior, 44(5), 1483–1492. https://doi.org/10.1007/s10508-014-0473-1. 52(4), 520–530. https://doi.org/10.1016/j.yhbeh.2007.07.013. Santtila, P., Antfolk, J., Ra ¨fsa ˚, A., Hartwig, M., Sariola, H., Sandnabba, Laeng, B.,Sirois, S.,&Gredeba ¨ck, G. (2012). Pupillometry: A window to the N. K., et al. (2015). Men’s sexual interest in children: One-year preconscious? Perspectives on Psychological Science, 7(1), 18–27. incidence and correlates in a population-based sample of Finnish https://doi.org/10.1177/1745691611427305. male twins. Journal of Child Sexual Abuse, 24(2), 115–134. https:// Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumu- doi.org/10.1080/10538712.2015.997410. lative science: A practical primer for t-tests and ANOVAs. Frontiers in Santtila, P., Mokros, A., Viljanen, K., Koivisto, M., Sandnabba, N. K., Psychology, 4, 863. https://doi.org/10.3389/fpsyg.2013.00863. Zappala, A., et al. (2009). Assessment of sexual interest using a choice Laws, D. R. (2009). Penile plethysmography: Strengths, limitations, innova- reaction time task and priming: A feasibility study. Legal and Crimi- tions. In D. Thornton & D. R. Laws (Eds.), Cognitive approaches to the nological Psychology, 14(1), 65–82. https://doi.org/10.1348/1355325 assessment of sexual interest in sexual offenders (pp. 7–30). Chichester, 07x267040. UK: Wiley. Savin-Williams, R. C., & Vrangalova, Z. (2013). Mostly heterosexual as Laws, D. R., & Gress, C. L. Z. (2004). Seeing things differently: The viewing a distinct sexual orientation group: A systematic review of the time alternative to penile plethysmography. Legal and Criminological empirical evidence. Developmental Review, 33(1), 58–88. https:// Psychology, 9, 183–196. doi.org/10.1016/j.dr.2013.01.001. Maletzky, B. M. (2003). Letter to the editor. Sexual Abuse: A Journal of Schmidt, A. F., Babchishin, K. M., & Lehmann, R. J. B. (2017). A meta- Research and Treatment, 15(4), 393. analysis of viewing time measures of sexual interest in children. McPhail, I. V., Hermann, C. A., Fernane, S., Fernandez, Y. M., Nunes, K. Archives of Sexual Behavior, 46(1), 287–300. https://doi.org/10. L., & Cantor, J. M. (2017). Validity in phallometric testing for 1007/s10508-016-0806-3. sexual interests in children: A meta-analytic review. Assessment. Schmidt, A. F., Banse, R., & Imhoff, R. (2015). Indirect measures in https://doi.org/10.1177/1073191117706139. forensic contexts. In T. M. Ortner & F. J. R. van de Vijver (Eds.), Mokros, A., Dombert, B., Osterheider, M., Zappala `, A., & Santtila, P. (2010). Behavior-based assessment in psychology: Going beyond self- Assessment of pedophilic sexual interest with an attentional choice report in the personality, affective, motivation, and social domains reaction time task.ArchivesofSexualBehavior,39(5), 1081–1090. https: (pp. 173–194). Go ¨ ttingen, Germany: Hogrefe. //doi.org/10.1007/s10508-009-9530-6. Scott, T. R., Wells, W. H., & Wood, D. Z. (1967). Pupillary response and Morokoff, P. J. (1986). Volunteer bias in the psychophysiological study sexual interest reexamined. Journal of Clinical Psychology, 23(4), of female sexuality. Journal of Sex Research, 22(1), 35–51. https:// 433–438. https://doi.org/10.1002/1097-4679(196710)23:4\433: doi.org/10.1080/00224498609551288. AID-JCLP2270230408[3.0.CO;2-2. Nunes, K. L., Firestone, P., & Baldwin, M. W. (2007). Indirect assessment Seto, M. C. (2008). Pedophilia and sexual offending against children: of cognitions of child sexual abusers with the Implicit Association Theory, assessment, and intervention. Washington, DC: American Test. Criminal Justice and Behavior, 34(4), 454–475. https://doi. Psychological Association. org/10.1177/0093854806291703. Seto, M. C. (2017). The puzzle of male chronophilias. Archives of Sexual O Ciardha, C. (2010). The use of implicit cognitive measures in the Behavior, 46(1), 3–22. https://doi.org/10.1007/s10508-016-0799-y. assessment of sexual offenders. Unpublished doctoral dissertation, Seto, M. C., & Lalumiere, M. L. (2001). A brief screening scale to identify Trinity College, Dublin, Ireland. pedophilic interests among child molesters.SexualAbuse:A Journal O Ciardha, C. (2011). A theoretical framework for understanding deviant of Research and Treatment, 13(1), 15–25. https://doi.org/10.1023/ sexual interest and cognitive distortions as overlapping constructs con- A:1009510328588. tributing to sexual offending against children. Aggression and Violent Singer, B. (1984). Conceptualizing sexual arousal and attraction. Behavior, 16(6), 493–502. https://doi.org/10.1016/j.avb.2011.05.001. Journal of Sex Research, 20(3), 230–240. O Ciardha, C., & Gormley, M. (2012). Using a pictorial-modified Stroop task Smith, P., & Waterman, M. (2004). Processing bias for sexual material: to explore the sexual interests of sexual offenders against children. Sex- The emotional Stroop and sexual offenders.SexualAbuse:AJournal ual Abuse: A Journal of Research and Treatment, 24(2), 175–197. of Research and Treatment, 16(2), 163–171. https://doi.org/10.1023/ https://doi.org/10.1177/1079063211407079. B:SEBU.0000023064.54397.d6. O Ciardha, C., & Gormley, M. (2013). Measuring sexual interest using a Snowden, R. J., & Gray, N. S. (2013). Implicit sexual associations in pictorial modified Stroop task, a pictorial Implicit Association Test heterosexual and homosexual women and men. Archives of Sexual and a choice reaction time task. Journal of Sexual Aggression, Behavior,42(3), 475–485. https://doi.org/10.1007/s10508-012-9920-z. 19(2), 158–170. https://doi.org/10.1080/13552600.2012.677486. Spiering, M., & Everaerd, W. (2007). The sexual unconscious. In E. Ortigue, S., & Bianchi-Demicheli, F. (2008). The chronoarchitecture of Janssen (Ed.), The psychophysiology of sex (pp. 166–184). Bloom- human sexual desire: A high-density electrical mapping study. Neu- ington, IN: Indiana University Press. Tanner, J. M. (1973). Growing up. Scientific American, 229(3), 34–43. 123 Arch Sex Behav (2018) 47:637–649 649 Thornton, D., & Laws, D. R. (2009). Cognitive approaches to the Wright, L. W., & Adams, H. E. (1994). Assessment of sexual preference assessment of sexual interest in sexual offenders. Chichester, UK: using a choice-reaction time task. Journal of Psychopathology and Wiley. Behavioral Assessment, 16(3), 221–231. Wolchik, S. A., Braver, S. L., & Jensen, K. (1985). Volunteer bias in erotica research: Effects of intrusiveness of measure and sexual background. Archives of Sexual Behavior, 14(2), 93–107. https:// doi.org/10.1007/bf01541656. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Archives of Sexual Behavior Pubmed Central

Latency-Based and Psychophysiological Measures of Sexual Interest Show Convergent and Concurrent Validity

Archives of Sexual Behavior , Volume 47 (3) – Dec 20, 2017

Loading next page...
 
/lp/pubmed-central/latency-based-and-psychophysiological-measures-of-sexual-interest-show-9E1ytd9R90

References (70)

Publisher
Pubmed Central
Copyright
© The Author(s) 2017
ISSN
0004-0002
eISSN
1573-2800
DOI
10.1007/s10508-017-1133-z
Publisher site
See Article on Publisher Site

Abstract

Arch Sex Behav (2018) 47:637–649 https://doi.org/10.1007/s10508-017-1133-z OR IG INA L PAP E R Latency-BasedandPsychophysiologicalMeasuresofSexualInterest Show Convergent and Concurrent Validity 1 2 1 • • Caoilte O Ciardha Janice Attard-Johnson Markus Bindemann Received: 9 January 2017 / Revised: 27 November 2017 / Accepted: 2 December 2017 / Published online: 20 December 2017 The Author(s) 2017. This article is an open access publication Abstract Latency-based measures of sexual interest require are available to researchers for this purpose (Savin-Williams & additional evidence of validity, as do newer pupil dilation approaches. Vrangalova, 2013). However, for research examining factors A total of 102 community men completed six latency-based mea- such as the cognitive mechanisms of sexual interests (e.g., sures of sexual interest. Pupillary responses were recorded during Ortigue & Bianchi-Demicheli, 2008), it becomes important three of these tasks and in an additional task where no participant to be able to measure those interests less directly, or even implic- response was required. For adult stimuli, there was a high degree itly. Additionally, there are applied contexts where understand- of intercorrelation between measures, suggesting that tasks may ing individuals’ sexual interests is clinically important but poten- be measuring the same underlying construct (convergent valid- tially vulnerable to social desirability biases—for example in the ity). In addition to being correlated with one another, mea- case of individuals who may have committed sexual offenses sures also predicted participants’ self-reported sexual inter- (see Kalmus & Beech, 2005). For this reason, much of the basic est, demonstrating concurrent validity (i.e., the ability of a task research that has examined the validity of indirect measures of to predict a more validated, simultaneously recorded, measure). sexual interest has done so with a distal or proximal goal of Latency-based and pupillometric approaches also showed pre- improving measurement of sexual interest relevant to forensic liminary evidence of concurrent validity in predicting both self- contexts, especially relating to sexual offending against chil- reported interest in child molestation and viewing pornographic dren. In the current study, we examined the validity of multiple material containing children. Taken together, the study findings methods of assessing sexual interest among male participants, build on the evidence base for the validity of latency-based again with the distal goal of improving forensic assessments, and pupillometric measures of sexual interest. but with relevance for broader sexology research. Despite perceptions to the contrary, not all individuals who Keywords Sexual interest  Viewing time  Reaction time  commit child sexual abuse have a sexual interest in children Pupil dilation  Pedophilia (Seto, 2008). While it may not be a universal treatment need, sexual interest in children is a leading predictor of sexual reof- fending among those convicted of a sexual offense (Hanson & Introduction Morton-Bourgon, 2004). Accurately assessing sexual interest is, therefore, vital for targeting criminogenic treatment needs For many research topics relating to sexuality and sexual behav- within rehabilitation and for estimating and managing risk. ior, it is sufficient for participants to self-report their sexual ori- The current ‘gold standard’ in assessing sexual interest is by mea- entation or sexual interests. A large number of self-report tools suring genital responses, typically using penile plethysmography (PPG). A recent meta-analysis by McPhail et al. (2017) found & Caoilte O Ciardha that PPG is a valid measure of pedohebephilic interest, but that C.C.OCiardha@kent.ac.uk validity varies depending on PPG methodology. Historically, PPG has faced questions of validity, reliability, and admissi- School of Psychology, Keynes College, University of Kent, Canterbury CT2 7NP, UK bility in court (for a review, see Laws, 2009). PPG is not rou- tinely used, for example, by Her Majesty’s Prison and Proba- Department of Psychology, Bournemouth University, Poole, tion Service in the UK in the assessment of individuals who UK 123 638 Arch Sex Behav (2018) 47:637–649 have committed sexual offenses. Where genital arousal is used While the IAT hinges on a facilitation effect, other reaction in sex research more broadly, there is a risk to the representa- time tasks depend on stimuli interfering with the primary task tiveness and generalizability of findings due to volunteer bias demands, thus suggesting that they hold some salience for the (Morokoff, 1986; Wolchik, Braver, & Jensen, 1985). Practi- participant. A variant of the choice reaction time (CRT) task tioners and researchers carrying out assessments of sexual inter- has been used where participants are instructed to report the est, therefore, need other indicators of sexual interest to comple- location of a dot on a screen while ignoring potentially sex- ment or provide an alternative to the use of PPG. ually appetitive stimuli on which that dot is overlaid. The As a response to the need for additional valid measures of CRT task typically results in longer reaction times to stimuli sexual interest, researchers have tested response latency-based consistent with participants’ sexual interests (e.g., Mokros, tasks (see Schmidt, Banse, & Imhoff, 2015; Thornton & Laws, Dombert, Osterheider, Zappala `, & Santtila, 2010; Wright & 2009). The most established task uses a viewing time paradigm Adams, 1994). Similarly, the modified Stroop task asks par- (Abel, Huffman, Warberg, & Holland, 1998) and is commer- ticipants to name the color of stimuli while ignoring their con- cially available as a technique for measuring sexual interest tent. Pictorial (O Ciardha & Gormley, 2012) and word (Price (for a meta-analysis of viewing time research, see Schmidt, & Hanson, 2007; Smith & Waterman, 2004) versions of the Babchishin, & Lehmann, 2017). Viewing time is based on the modified Stroop task have shown an interference effect of observation that individuals will spend longer looking at images stimulus content when that content is sexually salient. The they find sexually appealing than control or unappealing images. current study focuses on viewing time, IAT, CRT, and pictorial The task typically asks participants to rate images for attractive- modified Stroop (P-MST) tasks as latency-based measures of ness while simultaneously measuring response latency. Slowing sexual interest (for a broader review, see Schmidt et al., 2015). down of response times is interpreted as indicative of an indi- Compared to men, women—especially heterosexual women— vidual’s sexual interests. While the task is typically understood do not demonstrate the same level of category-specific respond- in relation to Singer’s (1984) theory of an aesthetic first phase in ing in physiological or latency-based tasks measuring sexual sexual arousal, the mechanisms underpinning this relatively interest (Snowden & Gray, 2013). For this reason, combined robust phenomenon are currently poorly understood (Imhoff with the lower prevalence rate of female-perpetrated sexual et al., 2010). There are concerns about the degree to which view- abuse, we focus only on the validity of tasks measuring sexual ing time tasks may be open to manipulation. Schmidt et al. (2017) interest among men. expressed surprise that there is a lack of research on the fakeability No single latency-based task currently holds a sufficient of viewing time. Additionally, for those undertaking assessment weight of evidence regarding its validity and reliability to be in adversarial settings, detailed task descriptions are available fully acceptable as a clinically useful measure of sexual inter- online; therefore, viewing time alone is not likely to provide a est. Indeed, case formulation is likely to be more reliably informed truly reliable measure of sexual interest for forensic assessment. by a confluence of clinically relevant information from multiple Other latency-based measures of sexual interest typically assessment methods rather than by identifying the single best demand that the participant carries out a task quickly and accu- performing task (Schmidt et al., 2017). Therefore, the question rately, while being presented with stimuli that may be related to of convergent and predictive/concurrent validity is of primary their sexual interests. Due to the fact that a rapid response is importance in building an evidence base for the clinical use of required in these tasks, we use the term reaction time tasks to measures of sexual interest. refer to them collectively and to differentiate them from view- Several studies have, to date, examined the convergent valid- ing time (Maletzky, 2003). In these tasks, sexually salient stimuli ity of different latency-based measures relating to sexual inter- may hinder or facilitate task performance in a way that is indica- est. Meta-analysis (Schmidt et al., 2017) based on a handful of tive of participants’ sexual interests. The Implicit Association studies has shown associations with small to medium effect Test (IAT) applied to the measurement of sexual interest asks sizes between viewing time and self-report, IAT, PPG, and the participants to categorize stimuli as quickly as they can use Screening Scale for Pedophilic Interests (Seto & Lalumie `re, 2001). two buttons (for a meta-analysis, see Babchishin, Nunes, & An earlier meta-analysis examining IAT research (Babchishin Hermann, 2013). In the trials of interest, each button has a et al., 2013) reported relationships with moderate effect sizes concept category and an attribute category assigned to it. If with viewing time, self-report, and the Screening Scale for the participant holds an association between the paired cat- Pedophilic Interests. egories, they should experience a facilitation effect when Ro ¨ nspies et al. (2015) compared viewing time with CRT, required to categorize stimuli. For example, Nunes, Firestone, and an adaptation of the Implicit Relational Assessment Pro- and Baldwin (2007) found that, when compared with other cedure (Barnes-Holmes et al., 2006) in measuring sexual ori- inmates, individuals who had committed sexual offenses against entation among male participants with exclusive sexual inter- children were quicker to categorize stimuli when paired cate- ests. The latter task is similar to the IAT but assesses an abso- gories were child/sexy and adult/not sexy than when the pairs lute rather than relative strength of relatedness/association were in the opposite arrangement. between concepts. Ro ¨ nspies et al. reported a convergence bet- 123 Arch Sex Behav (2018) 47:637–649 639 ween viewing time and the Implicit Relational Assessment Pro- respondence between pupillary responses and genital responses, cedure, both of which correlated with sexual orientation. The CRT particularly in males. Using non-nude images rather than videos, task used showed little evidence of construct validity. This was Attard-Johnson et al. (2016) found that straight men without a consistent with the findings of O Ciardha and Gormley (2013) self-reported sexual interest in children showed dilation to adult but not with the bulk of the published research which appears females but not to images of adult males or children. The current to demonstrate that the CRT can be used as a valid measure of study is the first to examine pupil dilation alongside multiple age-appropriate sexual interest or sexual interest in children latency-based measures of sexual interest. (Dombert et al., 2015; Gress, Anderson, & Laws, 2011;Mokros The study had two aims. First, we examined the convergent et al., 2010; Santtila et al., 2009;Wright & Adams, 1994). validity of viewing time, reaction time, and pupillometric mea- Using a community sample, the current study examined the sures of sexual interest in a non-offender sample. Participants convergent validity of viewing time (using three different sets performed a free-viewing task and three viewing time tasks (with of stimuli) alongside three reaction time tasks: a CRT task, a different stimuli) whereby pupil size was recorded, as well as P-MST, and a variant of the IAT. All tasks included both adult a CRT, a P-MST, and an IAT variant. The IAT variant (which and child stimuli. We tested concurrent validity by examining we refer to as a factorial IAT) included adult and child stimuli the relationship between all tasks and self-reported sexual inter- representing the concept categories of male and female.We ests in adults and some self-report indicators of possible sexual hypothesized that all measures would covary with one another, interest in children. No previous study has simultaneously com- demonstrating convergent validity, and that they would predict pared as many measures. Importantly, our study did not solely self-reported sexual interest, demonstrating concurrent valid- examine latency and self-report-based measures. We also incor ity. Second, we examined whether these tasks could also predict porated a psychophysiological measure of sexual interest into sexual interest toward children in this sample, further demon- the battery of tasks completed by participants. strating concurrent validity specifically in terms of their foren- The processes that facilitate or hinder speedy task perfor- sic application. This second aim was more exploratory given mance in reaction time tasks are typically seen as relatively the expectation of a low rate of sexual interest in children in a automatic and, therefore, less vulnerable to faking than other non-offender population. Santtila et al. (2015), for example, approaches. Where faking occurs, there is evidence that it is reported a one-year incidence of 3.3% for self-reported sexual detectable and may be correctable (Cvencek, Greenwald, Brown, interest in children among adult men. Gray, & Snowden, 2010). However, latency-based measures require input from the participant in the form of button-presses or vocal responding, which presents an opportunity for noise Method and/or deliberate manipulation to affect results. We, therefore, also included a psychophysiological method of assessing sex- Participants ual interest that may be more automatic than tasks requiring a response output from participants. Viewing content that is sex- One hundred and two men participated in this study in return ually appetitive elicits an instant automatic dilation of an for a small payment or course credit. Asamplesizetargetof100 observer’s pupils consistent with their sexual orientation (Attard- participants was set based on a combination of experience with Johnson, Bindemann, & O Ciardha, 2016; Rieger et al., 2015; previous experiments, pragmatic considerations (e.g., available Rieger & Savin-Williams, 2012). Importantly, pupil dilation funding), and a priori power analysis. We calculated the required appears highly resilient to deliberate manipulation (Laeng, sample size to test an overall multiple linear regression model Sirois, & Gredeba ¨ck, 2012). with ten predictors of sexual interest in adults with a medium The use of pupil dilation as a measure of sexual interest has and a large expected effect to equal 118 and 57 participants, a long history in the psychophysiology literature but has only respectively. Recruitment materials called for participants for recently achieved a level of consistency in findings that sup- a study ‘comparing eye tracking with other computer-based ports its validity. Using a somewhat rudimentary measurement measures of sexual interests.’ Participants were recruited using approach, Hess, Seltzer, and Shlien (1965) found evidence of a nontargeted approach via the university’s online recruitment pupil dilation to orientation-consistent stimuli among a small system, in addition to more purposive sampling through the uni- sample of gay and straight men. Other researchers (e.g., Scott, versity’s student LGBT? society social media page to increase Wells, & Wood, 1967), however, failed to replicate their gen- numbers of gay and bisexual men in the samples (for a similar eral findings. As a result, this method has been relatively under- approach, see Rieger & Savin-Williams, 2012). researched until recently, when Rieger and Savin-Williams (2012) found, using a large sample, that men’s pupils dilated to erotic video stimuli in line with their sexual interests. Rieger et al. (2015) further established the construct validity of pupil Attard-Johnson et al. (2017) presented a detailed examination of the dilation as a measure of sexual interest by demonstrating a cor- pupil dilation data of 100 participants from this data set. 123 640 Arch Sex Behav (2018) 47:637–649 Mean age was 22.2 years (SD = 5.7; range, 18–50 years). All Measures participants had completed secondary education. Ethnicity was not recorded in this study, but the sample was recruited through Pupil Dilation and Viewing Time Tasks advertising in a majority White but ethnically diverse university Participants completed four tasks (one free-viewing and three campus. Participants reported the degree to which they were attracted to male and female adults. The majority were exclu- viewing time tasks) while their pupil responses were recor- ded. Before commencing each task, the system was calibrated sively attracted to female adults (59.8%), 13.7% were predom- inantly attracted to females, 4.9% were attracted equally to using the standard nine-point fixation Eyelink procedure. male and female adults, 2.0% were predominantly attracted to male adults, and 19.6% were exclusively attracted to male adults. Free Viewing of Beach Set Stimuli In this task, participants All participants had normal or corrected-to-normal vision. were presented with the natural beach scenes. A free-viewing paradigm was adopted whereby participants were instructed to view the images as ‘naturally as they normally would’ (for Stimuli similar approaches, see Attard-Johnson et al., 2016; Hall, Hogue, &Guo, 2011). Each trial began with a fixation dot, which allowed Pictorial stimuli used across measures were from three sources. for drift correction, and ensured that participants attended to the TheNotRealPeople(NRP)set (clothed version; Laws & Gress, center of the display when the trial began. The trial began with a 2004) are composite images of males and females previously gray screen which was displayed for 1 s, followed by the stimulus developed for use in research involving sexual offenders. Images displayed for 10 s, followed by another gray screen for 1 s. Each in the NRP set are presented on a neutral background, classi- participant viewed all 25 images once with order randomized fied according to their secondary sexual characteristics (Tan- across participants. ner, 1973), and represent all five Tanner stages. A second set of images (Morph set;O Ciardha, 2010) provided a greater num- Viewing Time of Beach Set Stimuli The eye movement ber of adult (Tanner 5) stimuli than were available using the setup and procedure was identical to the previous task, except NRP set alone. These images are realistic but do not represent for the following. Scenes without person content were excluded real people, as faces are morphed composites of different faces. from the beach set stimuli. Instead of being instructed to view This set included control images of big cats. The third set of images ‘naturally as they normally would,’ participants were images (Beach set; Attard-Johnson et al., 2016) consisted of instructed to rate the sexual appeal of these targets on a seven- stock photographs portraying adult and prepubescent (Tanner point scale, using the number keys on a standard keyboard. 1) males and females in natural beach scenes (5 images in each Once a response was registered, the next trial was presented. of these four categories) and a set of control landscapes without any person content (5 scenes). People were portrayed in swim Viewing Time of NRP Set Stimuli This task employed the or leisure wear. same method and procedure as the previous viewing time task, except that the scene stimuli were replaced with images from the NRP picture set. A total of 40 images were used (four males Apparatus and four females at each Tanner stage). Pupil Dilation Tasks that included measurement of pupil Viewing Time of Morph Set Stimuli The procedure was dilation were displayed using SR-Research ExperimentBuilder identical to the previous viewing time tasks, except that a total software (version 1.1.0) on a 21’’ color monitor, with a screen of 36 computer-modified morph set stimuli were used (18 adult resolution of 10249 768 pixels. Eye movements were recorded males and 18 adult females). using an SR-Research Eyelink 1000 eye-tracking system. The Eyelink 1000 was running at a 1000-Hz sampling rate, a spatial Choice Reaction Time Task (CRT) resolution of\.01 of visual angle, a gaze position accuracy of\ .5, and a pupil size resolution of .1% of diameter. The Eyelink The CRT was presented using SuperLab4 stimulus presenta- 1000 system works by measuring corneal reflection and dark tion software and asked participants to identify as quickly as pupil with a video-based infrared camera. This system computes possible the location of a dot that appeared in one of five posi- the number of camera pixels that are occluded by partici- tions on an image of a person (or big cats in the practice con- pants’ pupils and records the area of the pupil as an integer that dition). Stimuli for this task, and for all other reaction time ranges from 400 to 16,000 units. During the recording of pupil tasks, were presented on a Dell E6520 laptop with a 15.6’’ color size, participants placed their heads on a chinrest approxi- monitor and a screen resolution of 13669 768 pixels. Responses mately 60 cm from the display monitor. were via a Cedrus RB-530 response pad. A white dot (approx- 123 Arch Sex Behav (2018) 47:637–649 641 imately 2.3 mm in circumference) was superimposed on each cross presented centrally for 500 ms. Participants could not con- of five positions on the images, yielding five versions of each tinue to the next trial until they had answered correctly. Stimuli image, where the white dot was located either in the top right, were presented at least twice within each block. top left, bottom right, bottom left, or middle of the image. Each The first f-IAT block presented stimuli, one at a time, from stimulus presentation was preceded by a centrally located fixa- the concept categories (adult and child images) and partici- tion cross for 500 ms. Stimuli remained on the screen until par- pants were instructed to identify them using the left and right ticipants responded. Participants received no feedback on errors, buttons on the response pad. Correct category headings were except for the first 20 practice trials. presented in the corresponding left and right corner of the Stimuli were taken from the NRP set and the morph set. screen. In the second block, participants categorized word Images were classified by age as pre-/early pubertal (Tanner 1 stimuli, again using the left and right buttons, as belonging to and 2), late pubertal (Tanner 3 and 4), and adult (Tanner 5), and one of the attribute categories. Sexual words were‘‘lust,’’‘‘lick,’’ by sex (female and male) to yield six trial types. Trial types were ‘‘kiss,’’‘‘naked,’’‘‘orgasm,’’‘‘arouse,’’and‘‘attractive,’’whilenon- presented together in small blocks or clusters of ten images. sexual words were ‘‘ugly,’’ ‘‘cold,’’ ‘‘dull,’’ ‘‘avoid,’’ ‘‘bland,’’ There were four clusters of each trial type, yielding 40 images ‘‘boring,’’and‘‘unattractive.’’In the third block, the left-hand con- of each trial type. Trials were randomized in each cluster, and cept category from block one was paired with the left-hand cluster order was randomized with the constraint that one of attribute category from block two (e.g., male and sexual), and each trial type cluster had to be presented before seeing a second likewise for the right-hand concept and attribute categories (e.g., of any type and so on. Response times and response correctness female and nonsexual). Only adult images were presented. The were recorded. fourth block was identical to the third except that only child images were presented. Pictorial Modified Stroop Task (P-MST) In the fifth block, the order of the concept pairs was swapped, whereby female might now appear on the left and male on the The P-MST asked participants to identify as quickly as possible, right. Both adult and child images were included in this block. by pressing colored buttons on the response pad, the color in In blocks six and seven, participants categorized both images which an image was presented, again using SuperLab4. Images and words by concept or attribute but with this new arrange- from the NRP and the morph sets had been tinted red, green, ment of the concept pairs. In block six, images were of adults, blue, and yellow. Four colored versions of twelve images each and in block seven images were of children. were used to create blocks of child male, child female, adult Participants’ response times to blocks 3/4 and 6/7 were ana- male, and adult female images, resulting in 48 images in each lyzed as differences between them should indicate differences block. Unlike the CRT, blocks of child stimuli combined images in the relative strength of associations between the concepts of male and female and the attributes of sexual and nonsexual. rated as belonging to Tanner stages 1–4. At the beginning of the task, participants completed a block of 16 practice trials fea- Furthermore, the strength of these associations could be exam- turing pictures of big cats. Feedback was given for incorrect ined to explore the impact of child stimuli. Order was random- responses for practice trials but not for the experimental trials. ized for participants so that some were presented with male Presentation of stimuli was preceded by a centrally located fix- sexual associations first, whereas others were presented first ation cross for 500 ms. Stimuli remained on the screen until with female-sexual pairings. Blocks three and six contained participants responded. A blocked design was adopted, whereby adult stimuli for all participants, with child stimuli in blocks all adult male images were presented together, as were all adult four and seven. females, and so on. Order of block presentation was randomized. As with the CRT, response times and response correctness were Interest in Child Molestation (ICM) Scale and Problematic recorded. Sexual Behavior or Fantasy Factorial Implicit Association Test (f-IAT) The scenarios used in this study were identical to the original ICM scale developed by Gannon and O’Connor (2011), with minor The f-IAT examined the strength of association between male/ changes to question order and wording. The scale presented five female concept categories and sexual/nonsexual attribute cate- scenarios describing sexual abuse with a child (age and gender gories. Stimuli for the concept category were images from the unspecified). Participants were asked to imagine themselves in NRP and morph sets. The f-IAT was based on an IAT used by the situation described. They were then asked to answer the O Ciardha and Gormley (2013), but included images of adults following questions on a seven-point scale: (1)‘‘In this situation, and children resulting in an extra‘‘factorial’’element. Twelve how sexually aroused would you be?’’(anchor points on the scale male and 12 female images were used, half of adults (Tanner 5) were ‘‘not at all sexually aroused’’ and ‘‘very strongly sexually and half of children (Tanner 1–3). The f-IAT was administered aroused’’); (2) ‘‘In this situation, how much would you enjoy over seven blocks. Each stimulus was preceded by a fixation getting your way?’’ (anchors of ‘‘would not enjoy it at all’’ and 123 642 Arch Sex Behav (2018) 47:637–649 ‘‘would greatly enjoy it’’; and (3) ‘‘If you found yourself in a 25th and 75th percentiles within that experimental condition similar situation, would you have done the same?’’ (anchors of for that individual. All reaction time data were converted to ‘‘would definitely not have done the same’’and‘‘would definitely ipsative z-scores. Ipsative z-scores present an individual’s have done the same’’). The total possible score on the ICM scale mean response times for a given condition in terms of numbers ranged from 15 to 105, with the minimum score indicating an of standard deviations (based on their own data) from their over- emphatic rejection of arousal, enjoyment, or propensity across all all mean response time. Therefore, if their response times on a child molestation conditions. P-MST to images of adult females yields an ipsative z-score Additional questions were included to examine prior sex- of ? 2, they have been responding slower to images of adult ual offending behavior and sexual fantasy involving children females than their overall mean reaction time by a magni- among participants. They were asked if they (1) had ever know- tude of two standard deviations. ingly and deliberately viewed pornographic material contain- Calculating ipsative z-scores for the f-IAT differs from the ing individuals below the age of consent, (2) had, since the age more standardD algorithm approach, suggested by Greenwald, of 18, found the thought of sex or sexual contact with a person Nosek, and Banaji (2003). However, this alternative has been 15 years of age or younger exciting or arousing, (3) had sexual used previously with a similar IAT (O Ciardha & Gormley, contact with a child under 13 who was more than 5 years younger 2013) and has the strength of being directly comparable with than them since the age of 16, or (4) had sexual contact with an the approach taken with the P-MST and CRT tasks. Addition- individual aged 13–15 who was more than 5 years younger since ally, the ipsative z-score approach allowed the separate exam- the age of 18. Only the results of question 1 are presented in the ination of the blocks in which adult and child stimuli were current study. presented. Procedure Viewing Time Tasks The study was reviewed and approved by the School of Psychol- As with the reaction time data, viewing time task response times ogy Research Ethics Committee (REF 20122341). Participants were converted to ipsative z-scores. No outlier removal was car- were seated in a windowless room with constant artificial light- ried out as there were fewer experimental trials in each condition ing. Participants completed tasks involving pupil dilation first compared with the reaction time tasks, and participants were not and in the following order: free viewing of beach set, viewing instructed to respond rapidly, meaning even conservative outlier time of beach set, viewing time of NRP set, and viewing time of removal risks removing trials that are not inconsistently slow in morph set. Participants then completed the reaction time tasks the context of the task instructions. (CRT, P-MST, and IAT). The order of reaction time tasks was counterbalanced. Finally, participants completed a brief ques- Pupil Dilation tionnaire to assess sexual interests followed by the ICM scale and questions about problematic sexual behavior or fantasy. In Observers’ pupillary responses to each stimulus category were a small number of cases, participants did not complete partic- calculated as a percentage change from observers’ overall pupil ular tasks or data did not record properly for technical reasons. mean. For this, pupillary responses were first computed by tak- We report sample sizes for individual analyses to reflect this. ing the mean pupil area at each fixation, averaged across the The average length of the entire procedure including obtaining duration of a stimulus display. An overall mean, across all stim- consent, questionnaires, completion of indirect tasks, calibration uli in all conditions, was then computed from these values for and administration of eye tracking/pupil dilation paradigms, and each participant. The percentage difference (i.e., an increase or debriefing was approximately 90 min per participant. decrease) in pupil area for each stimulus category (e.g., adult males) from the overall mean was then computed, using the Data Preparation formula: (mean pupil area for category 9 100)/overall pupil mean. Accordingly, a score of 100% indicates that the pupillary Reaction Time Tasks response to a stimulus category does not differ from the overall mean. Scores above or below this value indicate comparatively Reaction time data were calculated based on correct responses larger or smaller pupil sizes (for similar approaches, see Attard- only and were trimmed of outliers by removing times more Johnson et al., 2016; Laeng & Falkenberg, 2007). extreme than three times the interquartile range beyond the 123 Arch Sex Behav (2018) 47:637–649 643 carry out preliminary investigations to examine whether par- Results ticipants who appeared to demonstrate sexual interest in younger Assessing Sexual Interest in Adults individuals—through increased attention, stronger sex associa- tions, or greater pupil dilation—also self-reported greater interest For each of the measures (reaction times, viewing times, and in child molestation or were more likely to report having previ- ously used pornographic material containing children. pupil dilation), an index of preferred adult sex was calculated by subtracting the mean responses to adult female stimuli from We coded our predictor variables into binary values indicat- ing whether individuals’ strongest average response was to adult the mean responses to adult male stimuli (see Table 1). Each of these ten indices were positively correlated, with effect sizes stimuli (Tanner 5) or to younger stimuli. Table 4 shows the percentage of participants with a stronger response to younger ranging from small to large (Cohen, 1988; see Table 2). Sexual stimuli across all tasks alongside rank-biserial correlations with interest measured as a five-point continuous variable (exclu- the ICM. The binary predictors of adult or child interest were sively straight, given a value of 1, to exclusively gay, given a entered into a multiple linear regression predicting ICM scores value of 5) positively correlated with the index of preferred adult gender for each measure (rs .39–77; see Table 2). (see Table 5). Overall, the model accounted for 21% of the vari- ance in ICM scores. This reflects a moderate effect size. A multiple linear regression model incorporating all indices of preferred adult gender explained 75% of the variance in sex- In our sample, there were too few participants who admit- ted having viewed pornographic material containing children ual interest (see Table 3). This amount of variance explained equates to a large effect size. However, it also demonstrates to examine multiple predictors of that outcome in a logistic regression. Therefore, we examined the relationship between that there was still some error in the ability of the indices of indirect measures and use of pornographic material contain- preferred gender to predict self-reported sexual interest. We ing children using a t test. The dependent variable in this case explored whether a source of error might stem from cases was the total number of tasks showing a stronger response to where individuals reported non-exclusive sexual interests, child stimuli compared to adults. Across the eight paradigms, by examining whether exclusivity moderated the relation- individuals who admitted use of pornographic material con- ship between predicted values in the regression model and observed sexual interest (all variables standardized), using taining children had a greater number of strongest responses to child stimuli (M = 3.8, SD = 1.5) compared with individuals the PROCESS macro for SPSS (model 1; Hayes, 2013). The analysis (n = 96) showed that the relationship between pre- whoreported nouse (M= 2.8, SD= 1.2). This difference,- .99, BCa 95% CI [- 1.91, - .06], represented a large effect size, dicted and observed values was weaker for those with non- exclusive sexual interests, b = .32; SE = .18; BCa 95% CI albeit with a wide confidence interval; t(94)=- 2.57, d = .80 (effect size calculated using supplementary materials from [- .04, .67], compared with exclusive sexual interests, b = .91; SE = .05; BCa 95% CI [.80, 1.01]. Lakens, 2013). Assessing Sexual Interest in Children Discussion The mean interest in child molestation (ICM) score for this sample was 20.6 (SD = 9.3; range 15–59), with just over half In our sample, indices of sex preference to adult stimuli across tasks were consistently related to one another and to self-reported the sample (51%) emphatically rejecting an interest in child sexual abuse (i.e., scoring 15). Additionally, 11.8% (n = 13) sexual interest toward adults, consistent with our hypotheses. All correlations between indices were positive though effect sizes admitted knowingly and deliberately viewing pornographic material containing individuals below the age of consent. varied from small to large. Correlations between indices of sex preference and self-reported sexual interest toward adults were Despite non-minimal responding in the ICM and admission again positive, with effect sizes ranging from moderate to of the use of pornographic material containing children by large. These relationships demonstrated evidence of the con- some participants, it is unlikely that the sample would contain vergent and concurrent validity of latency-based tasks and of enough individuals with a clear sexual interest in children to robustly examine whether the tasks were valid measures of pre- pupil dilation. ferred age as well as preferred sex. However, it was possible to 2 3 For reasons of parsimony, we used the word preferred with reference Some previous studies combine Tanner 4 and 5 images as representing to gender or age in describing whether response time or pupil dilation mature (e.g., Ebsworth & Lalumiere, 2012) or have omitted Tanner 4 tasks indicated greater sexual interest to stimuli of a particular sex or age images entirely (e.g., Banse, Schmidt, & Clarbour, 2010). We were unable across individuals. We used the word to indicate the relative strength of to isolate Tanner 4 images across all paradigms presented. We therefore associations, pupil dilations, etc., to different stimuli, not to imply volun- examined only whether individuals responded most strongly to Tanner 5 tary choice in sexual orientation (see Committee on Lesbian and Gay versus all other Tanner stages. We examine implications in the Discussion Concerns, American Psychological Association, 1991). section. 123 644 Arch Sex Behav (2018) 47:637–649 Table 1 Samples sizes, grand means, standard deviations, and ranges for indices of sexual interest in adults nM (SD) Range Min Max Reaction time measures P-MST 98 - .12 (.63) - 1.76 1.68 CRT 100 - .17 (.80) - 3.33 2.57 f-IAT 100 - .32 (.67) - 1.63 1.31 Viewing time tasks NRP 102 - .81 (1.34) - 3.21 2.86 Morph 100 - .41 (.96) - 1.88 1.64 Beach 100 - .39 (.93) - 2.21 1.54 Pupil dilation NRP 100 .27 (6.65) - 15.93 17.85 Morph 100 - 1.68 (5.18) - 18.14 8.46 Beach (VT) 100 - 2.6 (7.14) - 18.00 14.77 Beach (free) 100 - 3.51 (6.48) - 19.13 14.69 P-MST, pictorial modified Stroop task; f-IAT, factorial Implicit Association Test; NRP, not real people image set (Laws & Gress, 2004); VT, viewing time; reaction and viewing time values represent the mean difference in ipsative z-scores between responses to adult male versus adult female stimuli; pupil dilation values represent the mean difference between pupillary responses to adult male versus adult female stimuli as a percentage of individual mean pupil size. For all measures, negative means suggest greater sexual interest in female adults Combined indices of sex preference to adult stimuli predicted Taken together, and interpreted in the context of a low base rate 75% of the variance in self-reported sexual interest in adults. of pedophilia and hebephilia in community samples, these While some predictors did not uniquely contribute to the results are very preliminary indicators of concurrent validity model’s overall ability to predict self-reported sexual interest, of the tasks as measures of sexual interest in children. those that did spanned reaction time, viewing time, and pupil Our findings offer evidence that indirect measures of sexual dilation approaches. In a post hoc analysis, we found that exclu- interest operate predictably and in a manner that is consistent sivity of sexual interest moderated the strength of the relation- with one another, particularly with adult stimuli. This was true ship between the predicted values and the observed value in our of the cognitive and psychophysiological paradigms exam- regression model above. This finding appears consistent with ined in the current study, at least in the conditions experienced the finding by Rieger et al. (2015) of weaker correspondence by our participants. Different processes relating to sexual atten- between indices of arousal in bisexual men compared to gay tion or arousal are likely to underpin these effects depending on or straight men. the task (O Ciardha, 2011). Pupil dilation, for example, is likely Our examination of measures as predictors of sexual inter- to reflect increased activity in the sympathetic nervous system est in children was exploratory given the expectation of low (Bradley, Miccoli, Escrig, & Lang, 2008). IAT, on the other numbers of participants with sexual interest in children. To hand, measures the strength of association between simple con- determine whether indirect task performance suggested a cepts held in memory and therefore possibly reflects schematic sexual interest in children, we created post hoc dichotomous relationships in ‘implicit sexual memory’ (for a discussion of variables indicating whether the strongest response on any explicit and implicit sexual memory, see Spiering & Everaerd, given indirect measure was to child or adult stimuli. Table 5 2007). shows the percentage of individuals showing a strongest response The CRT and P-MST appear to be measures of selective to child stimuli across indirect measures, and how measures cor- attention. However, both these tasks in the current study did relate with ICM scores. It is noteworthy that the percentage not randomly present individual trials but rather combined of participants showing a stronger response to children varied stimuli of the same trial type into blocks or clusters of stimuli. considerably depending on the task. This is a point we return to This follows recommendations for maximizing modified Stroop later. Indirect tasks performance, dichotomized in this way effects with appetitive stimuli (specifically addiction Stroop explained 21% of the variance in ICM scores, corresponding to tasks, see Cox, Fadardi, & Pothos, 2006). With a blocked design, a moderate effect size. Knowingly and deliberately viewing it is not possible to examine whether differences in latencies are due to an instantaneous capturing of attention or because of pornographic material containing children was related to hav- ing stronger response to children in a greater number of tasks. carryover effects once a given stimulus has been erased. As a 123 Arch Sex Behav (2018) 47:637–649 645 Table 2 Correlation coefficients between indices of preferred adult sex and self-reported adult sexual interests 1 2345 678910 Reaction time measures 1. P-MST 2. CRT r .62 n 98 BCa 95% CI [.44, .74] 3. f-IAT r .49 .52 n 98 100 BCa 95% CI [.33, .61] [.40, .64] Viewing time tasks 4. NRP r .49 .43 .53 n 98 100 100 BCa 95% CI [.38, .59] [.29, .56] [.38, .66] 5. Morph r .55 .40 .50 .83 n 96 98 98 100 BCa 95% CI [.41, .66] [.26, .53] [.31, .65] [.76, .89] 6. Beach r .43 .37 .40 .67 .72 n 96 98 98 100 100 BCa 95% CI [.20, .61] [.11, .58] [.21, .57] [.55, .77] [.61, .81] Pupil dilation 7. NRP r .42 .37 .40 .64 .65 .48 n 96 98 98 100 100 100 BCa 95% CI [.28, .53] [.22, .53] [.23, .54] [.49, .75] [.52, .76] [.32, .62] 8. Morph r .44 .46 .53 .68 .68 .55 .66 n 96 98 98 100 100 100 100 BCa 95% CI [.31, .57] [.33, .60] [.39, .65] [.57, .77] [.57, .77] [.41, .68] [.50, .77] 9. Beach (VT) r .14 .29 .22 .39 .43 .45 .32 .49 n 96 98 98 100 100 100 100 100 BCa 95% CI [- .04, .32] [.04, .52] [.03, .40] [.23, .53] [.28, .57] [.30, .59] [.13, .49] [.34, .62] 10. Beach (free) r .15 .20 .23 .30 .39 .31 .29 .39 .38 n 96 98 98 100 100 100 100 100 100 BCa 95% CI [- .002, .31] [.02, .39] [.05, .40] [.11, .48] [.24, .53] [.13, .46] [.11, .46] [.23, .54] [.17, .54] Sexual orientation r .59 .46 .65 .72 .77 .58 .58 .69 .39 .44 n 98 100 100 102 100 100 100 100 100 100 BCa 95% CI [.46, .69] [.32, .58] [.52, .75] [.58, .83] [.67, .85] [.44, .69] [.44, .70] [.59, .77] [.24, .52] [.29, .59] P-MST, pictorial modified Stroop task; f-IAT, factorial Implicit Association Test; NRP, not real people image set (Laws & Gress, 2004); VT, viewing time; BCa 95% CI, 95% confidence intervals based on 5000 resamples using the bias-corrected and accelerated (BCa) bootstrap method result, while the current study provided evidence of criterion among our sample (83%) was the pupil dilation method using validity across this battery of tasks by demonstrating their con- NRP stimuli. The tasks using NRP stimuli were the only tasks vergence, research examining their discriminant validity is which generated response data for each Tanner stage individ- needed to better understand the processes underpinning indi- ually. Inspection of the raw data showed that the percentage vidual tasks. was driven in large part by the number of participants (31%) In the current study, tasks containing child images did not showing the greatest average pupil dilation to Tanner 4 stimuli. all have identical groupings of stimuli by stage of sexual devel- Ebsworth and Lalumie `re (2012) described Tanner 4 as repre- opment. This was a consequence of selecting tasks that had senting‘‘individuals who are on the cusp of reaching full sexual previously been piloted or otherwise tested within our labo- maturity’’ (p. 164). As a result, a sexual interest in Tanner 4 ratory as part of existing research streams. It is noteworthy that images may not be a particularly useful index of forensically the measure suggesting the highest rates of interest in children relevant sexual interest, especially given the low average age of 123 646 Arch Sex Behav (2018) 47:637–649 Table 3 Summary of multiple linear regression analysis examining Table 5 Summary of multiple linear regression analysis examining whether indices of preferred adult sex predict sexual orientation (on a whether indices of preferred age predict ICM scores five-point scale) B SE B b BCa 95% CI B SE B b BCa 95% CI Reaction time measures Reaction time measures P-MST 1.54 2.40 .08 [- 3.06, 6.74] P-MST .59 .22 .23 [.15, 1.03] CRT 1.79 1.83 .10 [- 1.71, 5.45] CRT - .22 .16 - .10 [- .54, .09] f-IAT 1.20 1.87 .06 [- 2.54, 4.88] f-IAT .55 .14 .23 [.25, .83] Viewing time tasks Viewing time tasks NRP - .97 1.86 - .05 [- 4.83, 2.67] NRP .22 .16 .19 [- .11, .56] Beach 10.32 5.46 .31 [.13, 22.00] Morph .44 .22 .27 [.001, .87] Pupil dilation Beach - .11 .13 - .07 [- .36, .12] NRP 5.80 1.37 .23 [3.53, 8.36] Pupil dilation Beach (VT) 2.28 2.66 .08 [- 2.10, 6.90] NRP .01 .02 .02 [- .04, .05] Beach (free) - 1.02 1.99 - .04 [- 5.23, 3.09] Morph .05 .03 .18 [- .01, .12] R .21 Beach (VT) - .002 .02 - .01 [- .04, .03] F 2.94 Beach (free) .03 .02 .14 [.002, .07] n 96 R .75 P-MST, pictorial modified Stroop task; f-IAT, factorial Implicit Asso- F 25.82 ciation Test; NRP, not real people image set (Laws & Gress, 2004); VT, n 96 viewing time; B, unstandardized regression coefficient; b, standardized regression coefficient; BCa 95% CI, 95% confidence intervals of B based P-MST, pictorial modified Stroop task; f-IAT, factorial Implicit Asso- on 5000 resamples using the bias-corrected and accelerated (BCa) ciation Test; NRP, not real people image set (Laws & Gress, 2004); VT, bootstrap method viewing time; B, unstandardized regression coefficient; b, standardized regression coefficient; BCa 95% CI, 95% confidence intervals of B based on 5000 resamples using the bias-corrected and accelerated (BCa) our participants. Despite this limitation, however, the pupil dila- bootstrap method tion method using NRP stimuli was a predictor of ICM scores, with inspection of descriptive data indicating that participants showing the strongest pupil dilation to Tanner 5 stimuli also had the lowest ICM scores. Future studies using indirect Table 4 Percentage of participants with a stronger average indirect task response to child stimuli and rank-biserial correlation between having a measures should be designed, where possible, to measure the stronger response to children and ICM scores relative strength of response to stimuli across the full range of n Percentage of Correlation with ICM sexual maturity. strongest response The most robust findings of our study were those estab- r BCa 95% CI rb to child stimuli (%) lishing convergent and concurrent validity across tasks mea- suring sexual interest using adult stimuli, particularly where Reaction time measures participants reported exclusive sexual interests in one sex or P-MST 98 29.6 .07 [- .14, .28] another. The more exploratory findings examining sexual CRT 100 56.0 .08 [- .12, .26] interest in children will require corroboration with future stud- f-IAT 100 44.0 .04 [- .16, .24] ies containing greater number of pedophilic and hebephilic par- Viewing time tasks ticipants. However, establishing criterion validity of sexual inter- NRP 102 40.2 - .02 [- .21, .18] est tasks for forensic use based on age-appropriate sexual interest Beach 100 8.0 .25 [.003, .44] using non-offending samples is an important step. Seto (2017), Pupil dilation for example, outlined a multidimensional view of sexual orien- NRP 100 83.0 .32 [.16, .45] tation, defining sexual orientation as a ‘‘stable tendency to pref- Beach (VT) 100 13.0 .10 [- .11, .30] erentially orient—in terms of attention, interest, attraction, and Beach (free) 100 14.0 .05 [- .15, .26] genital arousal—to particular classes of sexual stimuli’’ (p. 3). P-MST, pictorial modified Stroop task; f-IAT, factorial Implicit Asso- Seto (2017) argued that while sex is the usual dimension ciation Test; NRP, not real people image set (Laws & Gress, 2004); VT, along which sexual orientation is discussed, age, or more specif- viewing time; r , rank-biserial correlation coefficient; BCa 95% CI, rb ically physical and sexual maturity, represents a dimension on 95% confidence intervals based on 5000 resamples using the bias-cor- rected and accelerated (BCa) bootstrap method which people may hold a stable orientation. He referred to 123 Arch Sex Behav (2018) 47:637–649 647 Attard-Johnson, J., Bindemann, M., & O Ciardha, C. (2016). Pupillary variations in individual orientation on this age dimension as response as an age-specific measure of sexual interest. Archives of chronophilias, which include age preferences to prepubescent Sexual Behavior, 45(4), 855–870. https://doi.org/10.1007/s10508- children (pedophilia), pubescent children (hebephilia), young 015-0681-3. adults (teleiophilia), etc. Within this multidimensional model, Attard-Johnson, J., Bindemann, M., & O Ciardha, C. (2017). Heterosexual, homosexual, and bisexual men’s pupillary responses to persons at sexual interests may differ across individuals in terms of what is different stages of sexual development. Journal of Sex Research, attractive, but may not differ in how that attraction impacts atten- 54(9), 1085–1096. https://doi.org/10.1080/00224499.2016.1241857. tion and arousal (see also O Ciardha, 2011). Therefore, results Babchishin, K. M., Nunes, K. L., & Hermann, C. A. (2013). The validity when measuring teleiophilic sexual interest among men (e.g., of Implicit Association Test (IAT) measures of sexual attraction to children: A meta-analysis. Archives of Sexual Behavior, 42(3), 487– task validity) should generalize to the testing of male sexual inter- 499. https://doi.org/10.1007/s10508-012-0022-8. est toward other age groups. If they do not, there are implications Banse, R., Schmidt, A. F., & Clarbour, J. (2010). Indirect measures of for the conceptualization of pedophilia, for example, as dimen- sexual interest in child sex offenders: A multimethod approach.Crim- sionally related to a sexual interest in adults. inal Justice and Behavior, 37(3), 319–335. https://doi.org/10.1177/ Data were collected in the current study under non-adver- Barnes-Holmes, D., Barnes-Holmes, Y., Power, P., Hayden, E., Milne, R., sarial conditions. As a result, we cannot be certain which mea- & Stewart, I. (2006). Do you really know what you believe? Devel- sures are most resilient to faking or whether there are certain oping the Implicit Relational Assessment Procedure (IRAP) as a direct conditions under which some or all measures do not correspond measure of implicit beliefs. The Irish Psychologist, 32(7), 169–177. Bradley, M. M., Miccoli, L., Escrig, M. A., & Lang, P. J. (2008). The pupil as a to an individual’s sexual interest (i.e., discriminant validity). measure of emotional arousal and autonomic activation. Psychophys- However, we are least enthusiastic about the potential utility of iology, 45(4), 602–607. https://doi.org/10.1111/j.1469-8986.2008.006 the viewing time approach, as it is the most transparent method- 54.x. ology. Schmidt et al. (2017) state similar concerns and lament Cohen, J. (1988). Statistical power analysis for the behavioral sciences. New York, NY: Routledge Academic. the lack of fakeability studies. While knowledge of reaction Committee on Lesbian and Gay Concerns, American Psychological time taskssuchasthe P-MST, CRT, and f-IAT would theoret- Association. (1991). Avoiding heterosexual bias in language. Amer- ically also allow people to manipulate their responses, the fact ican Psychologist, 46(9), 973–974. https://doi.org/10.1037/0003-06 that rapid responses are required makes it less likely that par- 6X.46.9.973. Cox, W. M., Fadardi, J. S., & Pothos, E. M. (2006). The addiction-Stroop ticipants could produce a coherent but fake pattern of respond- test: Theoretical considerations and procedural recommendations. ing. Given that pupil dilation is an automatic response to activity Psychological Bulletin, 132(3), 443–476. https://doi.org/10.1037/ in the sympathetic nervous system (Bradley et al., 2008), it is 0033-2909.132.3.443. likely to be the task that is least susceptible to deliberate manipu- Cvencek, D., Greenwald, A. G., Brown, A. S., Gray, N. S., & Snowden, R. J. (2010). Faking of the Implicit Association Test is statistically detec- lation (Laeng et al., 2012). Future research should address issues table and partly correctable. Basic and Applied Social Psychology, of fakeability, discriminant validity, and of course include 32(4), 302–314. https://doi.org/10.1080/01973533.2010.519236. participants with an admitted sexual interest in children. How- Dombert, B., Antfolk, J., Kallvik, L., Zappala `, A., Osterheider, M., Mokros, ever, taken together our results are the first to demonstrate strong A., et al. (2015). Identifying pedophilic interest in sex offenders against children with the Indirect Choice Reaction Time Task. European Jour- evidence of the convergent and concurrent validity of viewing nal of Psychological Assessment. https://doi.org/10.1027/1015-5759/ time, reaction time, and pupillometric approaches to the mea- a000293. surement of sexual interest. Ebsworth, M., & Lalumie `re, M. L. (2012). Viewing time as a measure of bisexual sexual interest. Archives of Sexual Behavior, 41(1), 161– Acknowledgements This work was supported by a research grant from 172. https://doi.org/10.1007/s10508-012-9923-9. the National Organisation for the Treatment of Sexual Abusers (NOTA) Gannon, T. A., & O’Connor, A. (2011). The development of the Interest in to Caoilte O Ciardha and Markus Bindemann. Child Molestation Scale. Sexual Abuse: A Journal of Research and Treatment, 23(4), 474–493. https://doi.org/10.1177/1079063211412 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons. Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding org/licenses/by/4.0/), which permits unrestricted use, distribution, and repro- and using the Implicit Association Test: I. An improved scoring duction in any medium, provided you give appropriate credit to the original algorithm. Journal of Personality and Social Psychology, 85(2), author(s) and the source, provide a link to the Creative Commons license, and 197–216. https://doi.org/10.1037/0022-3514.85.2.197. indicate if changes were made. Gress, C. L. Z., Anderson, J. O., & Laws, D. R. (2011). Delays in attentional processing when viewing sexual imagery: The development and com- parison of two measures. Legal and Criminological Psychology, 18(1), 66–82. https://doi.org/10.1111/j.2044-8333.2011.02032.x. Hall, C., Hogue, T., & Guo, K. (2011). Differential gaze behavior towards References sexually preferred and non-preferred human figures. Journal of Sex Research, 48(5), 461–469. https://doi.org/10.1080/00224499.2010. Abel, G. G., Huffman, J., Warberg, B., & Holland, C. L. (1998). Visual reaction time and plethysmography as measures of sexual interest in Hanson, R. K., & Morton-Bourgon, K. (2004). Predictors of sexual recidi- child molesters. Sexual Abuse: A Journal of Research and Treat- vismanupdatedmeta-analysis. Ottawa, ON: Public Works and Govern- ment, 10(2), 81–95. https://doi.org/10.1023/A:1022063214826. ment Services Canada. 123 648 Arch Sex Behav (2018) 47:637–649 Hayes, A. F. (2013). Introduction to mediation, moderation, and con- roImage, 43(2), 337–345. https://doi.org/10.1016/j.neuroimage.20 ditional process analysis: A regression-based approach. New York, 08.07.059. NY: Guilford. Price, S. A., & Hanson, R. K. (2007). A modified Stroop task with sexual Hess, E. H., Seltzer, A. L., & Shlien, J. M. (1965). Pupil response of offenders: Replication of a study. Journal of Sexual Aggression, hetero- and homosexual males to pictures of men and women: A 13(3), 203–216. https://doi.org/10.1080/13552600701785505. pilot study. Journal of Abnormal Psychology, 70(3), 165–168. https:// Rieger, G., Cash, B. M., Merrill, S. M., Jones-Rounds, J., Dharmavaram, doi.org/10.1037/h0021978. S. M., et al. (2015). Sexual arousal: The correspondence of eyes and Imhoff, R., Schmidt, A. F., Nordsiek, U., Luzar, C., Young, A. W., & Banse, genitals. Biological Psychology, 104, 56–64. https://doi.org/10.1016/ R. (2010). Viewing time effects revisited: Prolonged response latencies j.biopsycho.2014.11.009. for sexually attractive targets under restricted task conditions. Archives Rieger, G., & Savin-Williams, R. C. (2012). The eyes have it: Sex and of Sexual Behavior, 39(6), 1275–1288. https://doi.org/10.1007/s10508- sexual orientation differences in pupil dilation patterns. PLoS ONE, 009-9595-2. 7(8), e40256. https://doi.org/10.1371/journal.pone.0040256. Kalmus, E., & Beech, A. R. (2005). Forensic assessment of sexual interest: A Ro ¨ nspies, J., Schmidt, A. F., Melnikova, A., Krumova, R., Zolfagari, A., review. Aggression and Violent Behavior, 10(2), 193–217. https://doi. & Banse, R. (2015). Indirect measurement of sexual orientation: org/10.1016/j.avb.2003.12.002. Comparison of the implicit relational assessment procedure, viewing Laeng, B., & Falkenberg, L. (2007). Women’s pupillary responses to sexually time, and choice reaction time tasks. Archives of Sexual Behavior, significant others during the hormonal cycle. Hormones and Behavior, 44(5), 1483–1492. https://doi.org/10.1007/s10508-014-0473-1. 52(4), 520–530. https://doi.org/10.1016/j.yhbeh.2007.07.013. Santtila, P., Antfolk, J., Ra ¨fsa ˚, A., Hartwig, M., Sariola, H., Sandnabba, Laeng, B.,Sirois, S.,&Gredeba ¨ck, G. (2012). Pupillometry: A window to the N. K., et al. (2015). Men’s sexual interest in children: One-year preconscious? Perspectives on Psychological Science, 7(1), 18–27. incidence and correlates in a population-based sample of Finnish https://doi.org/10.1177/1745691611427305. male twins. Journal of Child Sexual Abuse, 24(2), 115–134. https:// Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumu- doi.org/10.1080/10538712.2015.997410. lative science: A practical primer for t-tests and ANOVAs. Frontiers in Santtila, P., Mokros, A., Viljanen, K., Koivisto, M., Sandnabba, N. K., Psychology, 4, 863. https://doi.org/10.3389/fpsyg.2013.00863. Zappala, A., et al. (2009). Assessment of sexual interest using a choice Laws, D. R. (2009). Penile plethysmography: Strengths, limitations, innova- reaction time task and priming: A feasibility study. Legal and Crimi- tions. In D. Thornton & D. R. Laws (Eds.), Cognitive approaches to the nological Psychology, 14(1), 65–82. https://doi.org/10.1348/1355325 assessment of sexual interest in sexual offenders (pp. 7–30). Chichester, 07x267040. UK: Wiley. Savin-Williams, R. C., & Vrangalova, Z. (2013). Mostly heterosexual as Laws, D. R., & Gress, C. L. Z. (2004). Seeing things differently: The viewing a distinct sexual orientation group: A systematic review of the time alternative to penile plethysmography. Legal and Criminological empirical evidence. Developmental Review, 33(1), 58–88. https:// Psychology, 9, 183–196. doi.org/10.1016/j.dr.2013.01.001. Maletzky, B. M. (2003). Letter to the editor. Sexual Abuse: A Journal of Schmidt, A. F., Babchishin, K. M., & Lehmann, R. J. B. (2017). A meta- Research and Treatment, 15(4), 393. analysis of viewing time measures of sexual interest in children. McPhail, I. V., Hermann, C. A., Fernane, S., Fernandez, Y. M., Nunes, K. Archives of Sexual Behavior, 46(1), 287–300. https://doi.org/10. L., & Cantor, J. M. (2017). Validity in phallometric testing for 1007/s10508-016-0806-3. sexual interests in children: A meta-analytic review. Assessment. Schmidt, A. F., Banse, R., & Imhoff, R. (2015). Indirect measures in https://doi.org/10.1177/1073191117706139. forensic contexts. In T. M. Ortner & F. J. R. van de Vijver (Eds.), Mokros, A., Dombert, B., Osterheider, M., Zappala `, A., & Santtila, P. (2010). Behavior-based assessment in psychology: Going beyond self- Assessment of pedophilic sexual interest with an attentional choice report in the personality, affective, motivation, and social domains reaction time task.ArchivesofSexualBehavior,39(5), 1081–1090. https: (pp. 173–194). Go ¨ ttingen, Germany: Hogrefe. //doi.org/10.1007/s10508-009-9530-6. Scott, T. R., Wells, W. H., & Wood, D. Z. (1967). Pupillary response and Morokoff, P. J. (1986). Volunteer bias in the psychophysiological study sexual interest reexamined. Journal of Clinical Psychology, 23(4), of female sexuality. Journal of Sex Research, 22(1), 35–51. https:// 433–438. https://doi.org/10.1002/1097-4679(196710)23:4\433: doi.org/10.1080/00224498609551288. AID-JCLP2270230408[3.0.CO;2-2. Nunes, K. L., Firestone, P., & Baldwin, M. W. (2007). Indirect assessment Seto, M. C. (2008). Pedophilia and sexual offending against children: of cognitions of child sexual abusers with the Implicit Association Theory, assessment, and intervention. Washington, DC: American Test. Criminal Justice and Behavior, 34(4), 454–475. https://doi. Psychological Association. org/10.1177/0093854806291703. Seto, M. C. (2017). The puzzle of male chronophilias. Archives of Sexual O Ciardha, C. (2010). The use of implicit cognitive measures in the Behavior, 46(1), 3–22. https://doi.org/10.1007/s10508-016-0799-y. assessment of sexual offenders. Unpublished doctoral dissertation, Seto, M. C., & Lalumiere, M. L. (2001). A brief screening scale to identify Trinity College, Dublin, Ireland. pedophilic interests among child molesters.SexualAbuse:A Journal O Ciardha, C. (2011). A theoretical framework for understanding deviant of Research and Treatment, 13(1), 15–25. https://doi.org/10.1023/ sexual interest and cognitive distortions as overlapping constructs con- A:1009510328588. tributing to sexual offending against children. Aggression and Violent Singer, B. (1984). Conceptualizing sexual arousal and attraction. Behavior, 16(6), 493–502. https://doi.org/10.1016/j.avb.2011.05.001. Journal of Sex Research, 20(3), 230–240. O Ciardha, C., & Gormley, M. (2012). Using a pictorial-modified Stroop task Smith, P., & Waterman, M. (2004). Processing bias for sexual material: to explore the sexual interests of sexual offenders against children. Sex- The emotional Stroop and sexual offenders.SexualAbuse:AJournal ual Abuse: A Journal of Research and Treatment, 24(2), 175–197. of Research and Treatment, 16(2), 163–171. https://doi.org/10.1023/ https://doi.org/10.1177/1079063211407079. B:SEBU.0000023064.54397.d6. O Ciardha, C., & Gormley, M. (2013). Measuring sexual interest using a Snowden, R. J., & Gray, N. S. (2013). Implicit sexual associations in pictorial modified Stroop task, a pictorial Implicit Association Test heterosexual and homosexual women and men. Archives of Sexual and a choice reaction time task. Journal of Sexual Aggression, Behavior,42(3), 475–485. https://doi.org/10.1007/s10508-012-9920-z. 19(2), 158–170. https://doi.org/10.1080/13552600.2012.677486. Spiering, M., & Everaerd, W. (2007). The sexual unconscious. In E. Ortigue, S., & Bianchi-Demicheli, F. (2008). The chronoarchitecture of Janssen (Ed.), The psychophysiology of sex (pp. 166–184). Bloom- human sexual desire: A high-density electrical mapping study. Neu- ington, IN: Indiana University Press. Tanner, J. M. (1973). Growing up. Scientific American, 229(3), 34–43. 123 Arch Sex Behav (2018) 47:637–649 649 Thornton, D., & Laws, D. R. (2009). Cognitive approaches to the Wright, L. W., & Adams, H. E. (1994). Assessment of sexual preference assessment of sexual interest in sexual offenders. Chichester, UK: using a choice-reaction time task. Journal of Psychopathology and Wiley. Behavioral Assessment, 16(3), 221–231. Wolchik, S. A., Braver, S. L., & Jensen, K. (1985). Volunteer bias in erotica research: Effects of intrusiveness of measure and sexual background. Archives of Sexual Behavior, 14(2), 93–107. https:// doi.org/10.1007/bf01541656.

Journal

Archives of Sexual BehaviorPubmed Central

Published: Dec 20, 2017

There are no references for this article.