חומר רקע

PDF 49,284 תווים המסמך המקורי ↗

AGGRESSIVE BEHAVIOR Volume 34, pages 341–351 (2008) Pornography Use and Sexual Aggression: The Impact of Frequency and Type of Pornography Use on Recidivism Among Sexual Offenders Drew A. Kingston1, Paul Fedoroff2,3, Philip Firestone1, Susan Curry3, and John M. Bradford2,3 1School of Psychology, University of Ottawa, Ottawa, Ontario, Canada 2Royal Ottawa Health Care Centre, Integrated Forensic Program, Ottawa, Ontario, Canada 3Institute of Mental Health Research, University of Ottawa, Ottawa, Ontario, Canada : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : In this study, we examined the unique contribution of pornography consumption to the longitudinal prediction of criminal recidivism in a sample of 341 child molesters. We speciﬁcally tested the hypothesis, based on predictions informed by the conﬂuence model of sexual aggression that pornography will be a risk factor for recidivism only for those individuals classiﬁed as relatively high risk for re-offending. Pornography use (frequency and type) was assessed through self-report and recidivism was measured using data from a national database from the Royal Canadian Mounted Police. Indices of recidivism, which were assessed up to 15 years after release, included an overall criminal recidivism index, as well as subcategories focusing on violent (including sexual) recidivism and sexual recidivism alone. Results for both frequency and type of pornography use were generally consistent with our predictions. Most importantly, after controlling for general and speciﬁc risk factors for sexual aggression, pornography added signiﬁcantly to the prediction of recidivism. Statistical interactions indicated that frequency of pornography use was primarily a risk factor for higher-risk offenders, when compared with lower-risk offenders, and that content of porno- graphy (i.e., pornography containing deviant content) was a risk factor for all groups. The importance of conceptualizing particular risk factors (e.g., pornography), within the context of other individual characteristics is discussed. Aggr. Behav. 34:341–351, 2008. r 2008 Wiley-Liss, Inc. : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : Keywords: pornography; sexually explicit material; sex offender; recidivism INTRODUCTION The inﬂuence of pornography on sexual aggres- sion has been a longstanding question that has provoked considerable debate amongst profes- sionals working with sexual offenders [Lalumie` re et al., 2005; Seto and Eke, 2005; Seto et al., 2001]. Given the exponential growth and accessibility of pornographic material [Malamuth et al., 2000], evidence for or against a link between such material and aggression would be important to public policy debates and legislation, as well as the assessment and treatment of sexual offenders [Seto et al., 2001]. Pornography’s inﬂuence on sexually aggressive behavior has been examined at both the population and individual levels [Seto et al., 2001]. The former method explores the relationship between the avail- ability of pornography and ﬂuctuations in the incidence of sexual crimes [Kimmel and Linders, 1996; Kutchinsky, 1991], whereas the latter exam- ines the association between pornography use and the likelihood of sexual aggression within indivi- duals. Analyses at the population level have produced equivocal results [Court, 1976; Kutchins- ky, 1991] and several methodological problems have been identiﬁed [see Malamuth and Petipitan, 2007 for a review]. In brief, such criticisms have focused on the problematic approach of drawing conclusions about individuals from societies at large, in addition to the lack of emphasis placed on important moderator variables (e.g., cultural variability, type of pornography) and the inﬂuence these may have Published online 28 February 2008 in Wiley InterScience (www. interscience.wiley.com). DOI: 10.1002/ab.20250 Received 22 February 2007; Revised 5 November 2007; Accepted 5 November 2007 Grant sponsors: Social Sciences and Humanities Research Council of Canada Doctoral Fellowship; Ontario Graduate Scholarship. Correspondence to: Drew A. Kingston, School of Psychology, University of Ottawa, 120 University Private, Ottawa, Ontario, K1N 6N5, Canada. E-mail: [email protected] r 2008 Wiley-Liss, Inc. on individual susceptibility to sexually explicit material [see Malamuth et al., 2000; Malamuth and Huppin, 2005; Seto et al., 2001]. At the individual level, pornography’s presumed role in sexual aggression has been examined using analog measures of aggressive behavior (e.g., admin- istration of electric shock) and correlating self- reported use of pornography with ofﬁcial records of aggressive behavior [e.g., recidivism; Marshall, 1988; Seto and Eke, 2006]. In a meta-analysis of 33 studies (N 5 2,040), Allen et al. [1995a] examined the association between pornography and nonsexual aggression using pro- totypical analog measures of aggressive behavior. The analysis divided sexually explicit material into one of the following three categories: (a) nudity, (b) nonviolent sexual behavior, and (c) violent sexual behavior. Overall, results indicated an association between pornography and aggression. However, type of pornography was a moderator, such that exposure to nudity decreased aggression, whereas exposure to the latter two categories signiﬁcantly increased aggression. Within naturalistic settings, pornography’s inﬂu- ence on aggression has been explored in both offender and nonoffender populations [Hald et al., 2007; Seto and Eke, 2005, 2006]. In noncriminal populations, Malamuth et al. [2000] examined the relationship between frequency of pornography use and sexual aggression in a representative sample of men (n 5 2,972). Results indicated that pornography use was positively correlated with coercive sexual behavior and was predictive of sexual aggression. These ﬁndings have been supported in other studies demonstrating a signiﬁcant relationship between a higher frequency of pornography use and type of use (i.e., deviant images) with sexual aggression [e.g., Boeringer, 1994; Vega and Malamuth, 2007]. However, as noted below, follow-up analyses demonstrated that the association between porno- graphy and sexual coercion was largely based on those individuals assessed as high risk to offend sexually. In known groups of sexual offenders, pornogra- phy use has been assessed in terms of frequency and type of pornography used. However, much of this research has been equivocal. With regard to frequency of pornography consumption, for exam- ple, there has been some evidence suggesting that sexual offenders obtain and view more pornography [Abel, 1985; Marshall, 1988] than nonoffender control samples, whereas others have found either no difference between groups [Condron and Nutter, 1988], or that the comparison groups reported more pornography use than sexual offenders [Cook et al., 1971]. Pornography’s inﬂuence on sexual crime has also been examined in sexual offender populations, speciﬁcally. Indeed, a signiﬁcant proportion of offenders in studies described by Abel [1985] and Marshall [1988] reported being inﬂuenced to sexu- ally offend, as a result of viewing pornography. Interestingly, the type of pornography that was related to sexually aggressive behavior in Marshall’s study involved consensual depictions, suggesting that content of pornography may be less important with respect to sexual aggression [Marshall et al., 1991]. In fact, it has been suggested that individuals, particularly those demonstrating a propensity to- ward violence, may exhibit deviant fantasies, which can be elicited from various consensual depictions [Marshall, 1988, 2000]. However, of note, the speciﬁc content of pornography is often difﬁcult to elucidate, as such material may include content representing both ‘‘nondeviant’’ and ‘‘deviant’’ forms [Malamuth et al., 2000]. The research summarized above generally indi- cates that pornography consumption is associated with adverse behaviors under some conditions and complements additional controlled research that supports such an effect [see Malamuth et al., 2000; Vega and Malamuth, 2007, for summaries]. How- ever, it is clear that many individuals view porno- graphy and do not act out aggressively in interpersonal contexts. It has been suggested that the negative effects of pornography are associated with certain individuals based on the complex interaction with particular individual and cultural differences [Malamuth et al., 2000; Malamuth and Petipitan, 2007]. Relevant moderating variables may include, but are not limited to, a family background which fosters the development of inappropriate attitudes and schema involving women, as well as more proximal factors, such as transient emotional states (e.g., anger). It is these factors, for example, that place certain individuals at greater risk for experiencing a negative impact from pornography exposure. Of particular relevance to the current research, pornography’s inﬂuence on aggressive behavior has been examined within the context of pretest mea- sures of risk characteristics (i.e., risk to offend sexually and violently). Most of the research investigating the interaction effects between porno- graphy and other variables has been conducted under the organizational framework of the Hier- archical-Mediational Conﬂuence model [HMC; 342 Kingston et al. Aggr. Behav. Malamuth, 1986; Malamuth et al., 2000; Malamuth and Huppin, 2005; Vega and Malamuth, 2007]. In brief, the HMC model was constructed from research demonstrating that sexual aggressors pos- sess several key characteristics, which are present both developmentally and at the time of aggression. These predictor variables operationalize two pro- posed pathways to sexual coercion. The ﬁrst is hostile masculinity ,which refers to a constellation of personality traits, combining a hostile orientation, typically toward women and satisfaction obtained through dominating, humiliating, and controlling women. The second pathway is impersonal sex and describes a noncommittal, game-playing orientation toward sexual activity and describes individual differences in the willingness to engage in such acts without closeness or commitment [Malamuth, 1998, 2003; Malamuth et al., 1995]. As opposed to a path-oriented model, where the presence of a speciﬁc factor directly determines the criterion of interest, the HMC model provides both a cumulative and conditional-probability explana- tion for the causes of sexually aggressive behavior. In other words, the HMC model highlights the importance of investigating a particular predictor (e.g., pornography) within the context of other variables (e.g., pretest measures of risk character- istics) and this allows for the inclusion of relevant moderating variables in a predictive model. The relationship between pornography and sexual aggression has been investigated according to the conditional-probability approach suggested by the HMC model in noncriminal sexual aggressors [i.e., college students who self-report using sexual coercion; see Malamuth et al., 2000; Vega and Malamuth, 2007]. These investigators classiﬁed participants into varying levels of risk to behave in a sexually coercive manner, based on the HMC model’s dimensions described above, and examined the predictive utility of pornography use. Results indicated that pornography was a signiﬁcant additional predictor of sexual aggression, after controlling for the other risk factors described by the model and that frequency of pornography use was only a risk factor for individuals assessed to be ‘‘at relatively high risk’’ for perpetrating sexual aggression [Malamuth et al., 2000; Vega and Malamuth, 2007]. Speciﬁcally, this research high- lighted an interaction effect, in which individuals classiﬁed as low risk demonstrated a small associa- tion between frequency of pornography use and sexual aggression, whereas high-risk men showed a large effect between pornography and sexual aggres- sion. With respect to attitudes, Hald et al. [2007] obtained similar results, such that individuals assessed as low or moderate risk for aggression demonstrated no effect between frequency of por- nography use and negative attitudes toward women, whereas the highest risk group demonstrated a signiﬁcant relationship. The purpose of this study was to evaluate the role of pornography as a risk factor for aggression and to extend the ﬁndings of Malamuth and others— that is, to examine whether pornography use is a signiﬁcant predictor of sexual aggression, when moderated by general and speciﬁc risk character- istics. As such, we hypothesized that pornography use would be a risk factor for recidivism only for those individuals classiﬁed as relatively high risk for re-offending. This hypothesis was tested using the following three classiﬁcations of recidivism: (1) all criminal recidivism, (2) violent (including sexual) recidivism, and (3) sexual recidivism only (see below for operational deﬁnitions). In naturalistic settings, it is difﬁcult to operatio- nalize distinctions across type of pornography content, as it is difﬁcult to differentiate deviant and nondeviant forms of sexually explicit material. This has led some [see Malamuth et al., 2000; Vega and Malamuth, 2007] to focus on self-reported frequency of general pornography use (e.g., con- sumption of sexually explicit magazines, which has shown to be strongly correlated with the use of other types of pornography). As such, the main analyses in this study focused on frequency of pornography use, irrespective of the type of content. However, additional analyses were conducted to explore the relationship between the type of content (i.e., deviant pornography) and aggression. Currently, research pertaining to pornography use and aggression, moderated by individual risk factors, has utilized noncriminal populations (i.e., college students) and, as such, has neglected individuals with an ofﬁcial history of sexual coercion (i.e., sexual offenders). Moreover, most studies have predominantly used cross-sectional research designs, and thus, longitudinal data pertaining to the relationship between pornography and aggression have been noticeable limited. This paper addressed both of these limitations. METHOD Participants Participants were adult men who had been convicted of a hands-on sexual offence against an individual under the age of 16 at the time of the 343 Pornography Use and Sexual Aggression Aggr. Behav. offence (N 5 341). The average age of the sample was 39.6 years (range: 18–78; SD 5 12.0). The sample consisted of 211 (61.9%) intrafamilial child molesters and 130 (38.1%) extra-familial child molesters. The average education level of this sample was 11.61 years (SD 5 3.76 years). Of the 341 participants, 49% had previous charges or convictions for criminal offences, 32.8% had pre- vious charges or convictions for violent (including sexual) offences, and 23% had previous charges or convictions for sexual offences. The participants were assessed at a university teaching hospital in a large Canadian city between 1982 and 1992. If police records indicated that a participant had ever offended against an adult, they were excluded from the analyses. Participants with both related and unrelated victims were not avail- able in this database. All participants signed a consent form at the time of assessment permitting use of their data for research, which was conducted in compliance with the internal review board of the hospital. Portions of this sample have been exam- ined in other studies [see Firestone et al., 1999, 2006; Kingston et al., 2007; Nunes et al., 2002], but the relationship between pornography and recidivism was not investigated. Measures Static 99. The Static 99 [Hanson and Thornton, 1999] is a brief actuarial instrument designed to predict the long-term probability of sexual recidi- vism among adult male sexual offenders. The ten items were derived from the Rapid Risk Assessment for Sex Offence Recidivism [Hanson, 1997] and the Structured Anchored Clinical Judgment—Min [Grubin, 1998] and include earlier sex offences, earlier sentencing dates, noncontact sex offense convictions, nonsexually violent index offence con- victions, earlier nonsexual violent convictions, un- related victims, stranger victims, male victims, ever lived with a lover for 2 years, and age. The overall score is translated into one of the four risk categories: low (0,1); medium–low (2,3); medium–high (4,5); and high [6–12; Hanson and Thornton, 2000]. The Static 99 has demonstrated excellent interrater reliability in several studies [de Vogel et al., 2004; Doren, 2004; Hanson, 2001; Harris et al., 2003], as well as good concurrent validity [Roberts et al., 2002]. Recently, Hanson and Morton- Bourgon [2004] pooled the results of 21 studies (n 5 5,103 sexual offenders) and found the Static 99 to have moderate predictive accuracy for sexual recidivism (d 5 0.63) and violent recidivism (d 5 0.57). Scoring of the Static 99 adhered to the coding guidelines provided by Hanson and Thornton [1999] and was based on information gathered earlier to an individuals release date. However, there were some deviations from the coding rules and these have been outlined in detail elsewhere [see Nunes et al., 2002]. To provide an example, the item, ‘‘ever lived with a lover for 2 years’’ was coded based on the indication that the offender had cohabitated with an intimate partner (i.e., been married), irrespective of the amount of time the couple had lived together. As such, the measure is best described as a modiﬁed version. It should be noted that such modiﬁcations did not detract from this instrument’s predictive validity using a similar sample [Nunes et al., 2002]. Analyses were carried out using the risk cate- gories, ‘‘low,’’ ‘‘moderate-low,’’ and ‘‘moderate- high,’’ as described by Hanson and Thornton [1999]. The highest risk category (i.e., ‘‘high risk’’) was excluded from the analyses, given the small number of participants allocated to this category (n 5 4). Throughout this paper, these categories will be referred to as low, medium, and high, respectively. This reclassiﬁcation was done to avoid awkwardness and to compare risk levels within this sample. The reader should be aware that any reference to medium and high risk in this paper is equivalent to the moderate-low- and moderate- high-risk categories in the Static 99, respectively. The average score on the Static 99 for this sample was low (Mean 5 1.4; SD 5 1.65). Bradford Sexual History Inventory. Self-re- ported pornography use was collected at the time of assessment using the Bradford Sexual History Inventory [Bradford et al., 1987, 1991]. This inventory, which is completed by participants during an initial psychiatric interview, consists of 81 items grouped into nine categories and inquires about an individual’s sexual activity. For this study, questions pertaining to pornography use were of importance. Speciﬁcally, individuals were asked to rate the frequency with which they had viewed sexually explicit ﬁlms and/or books over the course of their lifetime. The corresponding response categories were 0, 1, 2–5, 6–10, 11–50, 51–100, 101–200, and more than 200. Each category was coded on a 1 to 8 scale with 8 representing the most frequent category (more than 200). Next, the individuals responded to a question concerning the type of pornography used and response categories were (1) heterosexual sex, (2) homosexual sex, (3) lesbian sex, (4) children engaged in sexual activity, and (5) depictions of 344 Kingston et al. Aggr. Behav. violence. Individuals could check more than one type of pornography used and deviance was deﬁned as any self-reported use of pornography containing children and/or violence. Recidivism analyses. The dependent measures in this study were organized in a cumulative hierarchical manner, beginning with a comprehen- sive category that included all types of recidivism, followed by more speciﬁc categories of recidivism. This classiﬁcation method is similar to other studies and thus, allows for comparison across studies [see Firestone et al., 1999, 2006; Rice et al., 1991]. Moreover, this cumulative hierarchical approach allows for the inclusion of sexually motivated offenses that were ‘‘pled down’’ to violent or criminal offences as many sexual offenders would rather admit to any offence other than a sexual offence. Evidence of recidivism was obtained from the Canadian Police Information Center’s report, a national database of criminal arrests and convictions from the Royal Canadian Mounted Police. The subcategories were as follows: (1) all criminal recidivism was used as the comprehensive overall measure of recidivism, which included any charge or conviction noted in the Canadian Police Information Center report (i.e., criminal, violent, and/or sexual); (2) violent (including sexual) recidivism was deﬁned as any charge or conviction of a violent and/or sexual offence (e.g., assault, assault causing bodily harm); and (3) sexual recidivism was deﬁned as any charge or conviction of a sexual offence (e.g., invitation to sexual touching). Speciﬁcally, the proportion of recidivists was calculated as a function of all new offences, regardless of when these offences occurred during the follow-up period. In this study, it should be stressed that recidivists are those men who have been charged or convicted of re-offending, and therefore these rates are approximations of true re-offense rates, as some men who committed these acts would not have been apprehended. The overall rates of recidivism in this study were 31.7% for criminal recidivism, 21.4% for violent recidivism, and 11.1% for sexual recidivism. The recidivism rates for the intrafamilial child molesters were 24.2%, 17.1%, and 8.1%, for criminal, violent, and sexual offences, respectively. The recidivism rates for the extra-familial child molesters were 43.8%, 28.5%, and 16.2%, for criminal, violent, and sexual offences, respectively. The follow-up period was assessed on release to the community and ranged up to 15 years, with an average of 8.4 years (SD 5 4.0 years). Statistical analyses. For this study, sequential logistic regression analyses were conducted to analyze the relationship between pornography use and recidivism and to address the possibility that an individual’s risk level would be a moderator of this relationship. To address the strength of the relationships in these analyses, Cohen’s d and odds ratios were reported. By convention, Cohen’s d effect sizes of .20, .50, and .80 are small, medium, and large, respectively [Cohen, 1988, 1992]. Additionally, 95% conﬁdence intervals (CI) around d were provided to indicate the range of values that would be expected in 95% of other samples utilizing the same popula- tion of sexual offenders. Values of d are considered statistically signiﬁcant if the 95% CI does not contain zero. Finally, when comparing effect sizes (see Fig. 1), values of d are signiﬁcantly different from one another when their 95% CIs are not overlapping. Odds ratios, as reported in the regres- sion analyses, can be interpreted as the increase or decrease in the predicted odds of recidivism, which corresponds to an increase of one point on the predictor variable (e.g., frequency of pornography use), or in the case of a dichotomous predictor (i.e., deviant pornography), the odds of recidivism in one group compared with the other. An odds ratio of 1 reﬂects no relationship between a predictor and a outcome. A series of sequential logistic regression analyses were conducted for each dependent measure to test the importance of the conditional-probability approach described by the HMC model in general and examining risk to re-offend, as a moderator between pornography use and recidivism, in particular. Speciﬁcally, porno- graphy and risk level were entered as independent variables. Consistent with Vega and Malamuth [2007], we divided pornography use into low, medium, and high frequency of use. Speciﬁcally, scores below the median (i.e., scores of 1–3), were assigned into the low- use category (n5116). Individuals scoring between 4 and 5 (n5147) were assessed as moderate users, and ﬁnally, the remaining individuals were assessed as high users (i.e., scores of 6–8; n578). RESULTS All Criminal Recidivism As can be seen in Table I, Static 99 score made a signiﬁcant contribution to the prediction of all recidivism (w2 change 5 15.61, df 5 1, Po.001), but frequency of pornography use did not (w2 change 5 .512, df 5 1, P 5 .474). The interaction 345 Pornography Use and Sexual Aggression Aggr. Behav. between risk level and pornography use was signiﬁcant (w2 change 5 6.57, df 5 1, Po.05), sug- gesting that the relationship between pornography use and recidivism was different across levels of risk. Violent (Including Sexual) Recidivism As can be seen in Table II, Static 99 risk level made a signiﬁcant contribution to the prediction of violent (including sexual) recidivism. In Block 2, the addition of pornography use made a signiﬁcant contribution to recidivism, after controlling for Static 99 risk level (w2 change 5 4.48, df 5 1, Po.05). In the third block, the pornography by Static 99 risk-level interaction was signiﬁcantly associated with the prediction of recidivism (w2 change 5 4.72, df 5 1, Po.05), suggesting that the relationship between pornography use and recidivism was different across levels of risk. Sexual Recidivism As indicated in Table III, Static 99 risk level made a signiﬁcant contribution to the prediction of sexual recidivism. Frequency of pornography use was added in Block 2 and did not make a signiﬁcant contribution to the prediction of sexual recidivism, after controlling for Static 99 risk level (w2 change 5 1.85, df 5 1, P 5 .174). The interaction between Static 99 and pornography use was also not signiﬁcant (w2 change 5 1.28, df 5 1, P 5 .259). 0.09 0.12 1.39 0.21 0.31 1.35 0.09 0.18 0.6 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 High Medium Low Static 99 Risk Levels (modified) Effect Size (d) Criminal Violent Sexual Fig. 1. Relationship between frequency of pornography use (continuous) and recidivism, as a function of risk to commit sexual aggression. TABLE I. Logistic Regression Analysis for Risk and Pornography Use Predicting Criminal Recidivism 95% CI for eb b SE eb Lower Upper w2 change from previous block Block 1 15.61 Static 99 .841 .219 2.34 1.51 3.56 Block 2 .512 Static 99 .820 .221 2.27 1.47 3.50 Pornography use .146 .204 1.16 .78 1.73 Block 3 6.57 Static 99 .811 .694 .444 .114 1.73 Pornography use 1.10 .544 .333 .115 .967 Interaction .828 .340 2.29 1.18 4.45 CI, conﬁdence interval. *Po.05; **Po.01; ***Po.001. 346 Kingston et al. Aggr. Behav. Interaction Between Pornography and Risk to Re-Offend The above analyses provided support for the hypothesis that propensity toward sexual aggression moderates the relationship between pornography use and aggression [see Malamuth et al., 2000; Vega and Malamuth, 2007]. To further examine these interactions, effect sizes were displayed across Static 99 risk categories (i.e., the redeﬁned low, medium, and high-risk categories) and examined with respect to frequency of pornography use (using the 1–8 scale). The effect sizes shown in Figure 1 highlight the interaction indicated in the previous analyses, such that individuals assessed as low risk (n 5 135) demonstrated small associations between criminal (d 5 .09, 95% CI 5 .15–.33), violent (d 5 .21, 95% CI 5 .03–.45), sexual (d 5 .09, 95% CI 5 .15–.33) recidivism, and the frequency of pornography use. Individuals assessed as medium risk (n 5 51) demonstrated small but elevated associations between frequency of porno- graphy and criminal (d 5 .12, 95% CI 5 .27–.50), violent (d 5 .31, 95% CI 5 .08–.69), and sexual (d 5 .18, 95% CI 5 .21–.57) recidivism. Finally, individuals assessed as high risk in our analysis (n 5 22) demonstrated moderate to large effect sizes between frequency of pornography use and criminal (d 5 1.39, 95% CI 5 .73–2.00), violent (d 5 1.35, 95% CI 5 .69–2.00), and sexual (d 5 .60, 95% CI 5 .01–1.20) recidivism. As evidenced by the CIs, there were signiﬁcant differences between individuals assessed as high risk and low risk for criminal and violent recidivism. Additional Analyses Regarding Pornographic Content Of the 341 child molesters in this study, 337 responded to questions pertaining to type of content. Among these individuals, 303 (90%) reported viewing only nondeviant pornography, TABLE III. Logistic Regression Analysis for Risk and Pornography Use Predicting Sexual Recidivism 95% CI for eb b SE eb Lower Upper w2 change from previous block Block 1 4.91* Static 99 .618 .272 1.85 1.09 3.16 Block 2 1.85 Static 99 .565 .276 1.76 1.03 3.02 Pornography use .288 1.48 .839 2.59 .388 Block 3 1.28 Static 99 .412 .935 .662 .106 4.14 Pornography use .351 .717 .704 .173 2.87 Interaction .452 .408 1.57 .707 3.50 CI, conﬁdence interval. *Po.05; **Po.01; ***Po.001. TABLE II. Logistic Regression Analysis for Risk and Pornography Use Predicting Violent (Including Sexual) Recidivism 95% CI for eb b SE eb Lower Upper w2 change from previous block Block 1 12.11 Static 99 .782 .225 2.19 1.41 3.40 Block 2 4.48 Static 99 .726 .229 2.07 1.32 3.24 Pornography use .479 .229 1.61 1.03 2.53 Block 3 4.72 Static 99 .838 .794 .433 .091 2.05 Pornography use .695 .599 .499 .154 1.61 Interaction .754 .366 2.13 1.04 4.35 CI, conﬁdence interval. *Po.05; **Po.01; ***Po.001. 347 Pornography Use and Sexual Aggression Aggr. Behav. whereas 34 (10%) indicated viewing deviant porno- graphy. The use of deviant pornography was unrelated to risk level (r 5 .07, P 5 .323). Given the few participants within the recidivist categories, caution is warranted when interpreting these results. Nevertheless, to highlight possible trends, a series of logistic regression analyses were conducted to test for possible interactions between the three-level hierarchical risk variable and the two-level type of content variable on the dependent measures. With regard to criminal recidivism, risk level made a signiﬁcant contribution to the prediction of recidi- vism in Block 1 (w2 change 5 14.36, df 5 1, Po.001). The addition of pornography content into the equation was signiﬁcant, after considering risk level (w2 change 5 4.99, df 5 1, Po.05). The odds ratio indicated that for individuals who viewed deviant pornography, the predicted odds of criminal recidi- vism increased by 177% when compared with those who did not view deviant pornography. The interaction between risk level and type of porno- graphy was not signiﬁcant (w2 change 5 .293, df 5 1, P 5 .588). In terms of violent (including sexual) recidivism, risk level made a signiﬁcant contribution to the prediction of recidivism in Block 1 (w2 change 5 11.62, df 5 1, Po.01). The addition of type of pornography was signiﬁcant, after control- ling for risk level (w2 change 5 4.94, df 5 1, Po.05). The odds ratio indicated that for individuals who viewed deviant pornography, the predicted odds of violent (including sexual) recidivism increased by 185% when compared with those who did not view deviant pornography. The interaction between risk level and type of pornography was not signiﬁcant (w2 change 5 .999, df 5 1, P 5 .317). Finally, both risk level (w2 change 5 4.46, df 5 1, Po.05) and pornography content (w2 change 5 4.83, df 5 1, Po.05) made signiﬁcant contributions to the pre- diction of sexual recidivism for Blocks 1 and 2, respectively. The odds ratio indicated that for individuals who viewed deviant pornography, the predicted odds of sexual recidivism increased by 233% when compared with those who did not view deviant pornography. The interaction between these variables was not signiﬁcant (w2 change 5 1.11, df 5 1, P 5 .292). DISCUSSION The purpose of this study was to examine the relationship between pornography and aggressive behavior within the context of an important moderating variable—that is, risk to re-offend [Malamuth, 2003]. According to recent investiga- tions, the predictive utility of pornography is based on the interaction between various risk character- istics associated with aggression [Malamuth et al., 2000], and individuals who view sexually explicit material are more likely to offend and/or re-offend when they possess such characteristics [Hald et al., 2007; Vega and Malamuth, 2007]. The results of this study supported the utility of pornography as a predictor of aggression, when examined in con- ﬂuence with other general and speciﬁc risk factors for aggression. We examined the impact of frequency of porno- graphy use on the overall comprehensive measure of criminal recidivism, as well as the more speciﬁc categories of violent (including sexual) recidivism and sexual recidivism only. Results indicated that the frequency of pornography use contributed to the prediction of criminal and violent recidivism, while taking other risk factors for sexual aggression into account. Follow-up analyses indicated that the interaction between pornography and risk to re- offend was consistent with the conditional-prob- ability model outlined in the HMC model. Speciﬁ- cally, we found that among men who scored high on general and speciﬁc risk characteristics, frequent pornography consumption increased the risk for aggression. In contrast, amount of pornography use had little predictive value for men assessed to be at low risk for sexual aggression. The predictive utility of pornography use among high risk, as opposed to low-risk individuals, has been explained by social learning theory in general, and the notion of reciprocal determinism, deﬁned as the interaction between person, behavior, and environment, in particular [Malamuth and Huppin, 2005; Seto et al., 2001]. Speciﬁcally, individuals with a predisposition for aggression (i.e., men who are at relatively high risk for aggression) have shown to be particularly drawn to images of pornography and are more likely to expose themselves in the future to such images than lower-risk individuals [Shim et al., 2007]. Moreover, a number of priming studies have shown that men with earlier risk characteristics may interpret sexually explicit material differently than lower-risk individuals, such that pornography acti- vates and reinforces inappropriate cognitive repre- sentations (e.g., hostility toward women) and fosters the development of sexual preoccupation in these men [see Malamuth et al., 2000 for a review]. Given that both of these factors are related to future sexual aggression [Hanson and Morton-Bourgon, 2004], it is not surprising that men who were assessed as relatively high risk for sexual aggression and who 348 Kingston et al. Aggr. Behav. were frequent users of pornography were more likely to behave aggressively compared with lower-risk offenders. Of note, the main effects and interactions between frequency of pornography use and sexual recidivism were not signiﬁcant. This was somewhat surprising given research suggesting that pornography use is associated with sexual coercion [e.g., Malamuth et al., 2000], and that the observed interactions between pornography use and risk to re-offend have been demonstrated for this type of behavior [Vega and Malamuth, 2007]. Importantly, however, the interaction was signiﬁ- cant for violent (including sexual) recidivism, which we feel is a better representation of the inﬂuence of pornography on sexually aggressive behavior. In fact, several investigators have argued that using violent (including sexual) recidivism is the most accurate outcome criterion when interested in sexual recidivism, given the tendency of the former to ‘‘capture signiﬁcantly more sexual re-offenses than the more commonly used sexual recidivism deﬁnition’’ [Quinsey et al., 1998, p. 129]. Recently, this was demonstrated empirically [Rice et al., 2006] in a comparison of 177 police rapsheets (ofﬁcial documentation regarding charges and convictions) with more detailed clinical case reports. Results indicated that approximately 33% of offenders for whom no sexual motivation was indicated had most likely committed a sexually motivated crime. The ﬁndings indicated by Rice et al. supports using violent (including sexual) recidivism as the most reliable outcome measure when interested in sexually motivated offences. As such, the signiﬁcant interaction found among violent (includ- ing sexual) recidivists in our study replicates and extends the ﬁndings of Malamuth et al. [2000] and Vega and Malamuth [2007] indicating that individual risk is an important variable moderating the relationship between pornography and sexual aggression. Next, we examined the degree to which self-reported use of deviant pornography was predictive of the overall comprehensive measure of criminal recidivism, as well as the more speciﬁc categories of violent (including sexual) recidivism and sexual recidivism. Results supported a main effect of pornographic content, after controlling for general and speciﬁc risk characteristics, as contained in the Static 99. Speciﬁcally, results indicated that individuals who viewed deviant pornography were more likely to recidivate when compared with individuals who did not view deviant pornography and this difference was consistent across levels of risk (i.e., no interactions). There is a growing body of literature investigating the impact of exposure to deviant pornography on attitudes supportive of sexual aggression [Allen et al., 1995a,b; Malamuth and Check, 1981], physiological arousal to sexual aggression [Mala- muth et al., 2000; Marshall et al., 1991; Seto et al., 2006], and actual aggressive behavior among nonoffenders [Malamuth et al., 2000]. Thus far, results have generally supported the negative impact from viewing deviant pornography on these out- come measures and our ﬁndings were consistent with such results. Both observational learning and conditioning processes suggest that repeated exposure to deviant forms of pornography, given the focus on male entitlement and power, help shape an individual’s fantasies, perceptions, rationalizations, and deeper core beliefs [Lalumie` re et al., 2005; Marshall, 2000; Seto et al., 2001]. It is important to note that such development is most likely multifaceted and that pornography may simply accelerate a process that is already underway [Marshall, 2000]. Of equal importance, however, is that the impact of deviant pornography on behavior was consistent across levels of risk. This suggests that exposure to unconventional sexual activity fosters the progression toward re- offending, regardless of the earlier existence of historical risk factors. In contrast, frequency of pornography use, as indicated above, was a pre- dictor for individuals already possessing such a predisposition toward re-offending. Several issues must be considered when interpret- ing these results. First, the assessment of pornogra- phy was problematic, as it was based solely on self- report and required individuals to recall information over the course of their life times. Regarding the ﬁrst point, individuals undergoing assessment in a forensic setting are sometimes reluctant to be forthcoming with information, especially when such information could have negative consequences for their evaluation. This type of limitation is consis- tently identiﬁed in forensic research [e.g., Nugent and Kroner, 1996]. Additionally, individuals were asked to recall information spanning much of their lifetime and problems with adequate retrieval of early events may have inﬂuenced the results. A ﬁnal problem regarding the assessment of pornography use pertained to the type of pornographic stimuli examined. In other words, the type of pornography involved ﬁlms and/or books and thus neglected Internet pornography. Unfortunately, given the dates of assessment (1982–1992), this was not 349 Pornography Use and Sexual Aggression Aggr. Behav. possible and future research should examine similar questions pertaining to individuals who use the Internet to obtain sexually explicit material. In spite of these limitations, this current research supported and extended the results reported by other studies with noncriminal sexual aggressors indicating that pornography exposure was a signiﬁcant predictor of aggression when examined in conﬂuence with other risk factors. Speciﬁcally, this study highlighted the importance of considering various interactive factors that can act synergisti- cally in determining the probability for a particular behavioral outcome. The important implications of the cumulative-conditional-probability concep- tualization, as described in research investigating the HMC [Malamuth et al., 2000], is not limited to pornography use but has important implications for examining the complex relationships between distal and proximal factors as predictors of sexual aggression. ACKNOWLEDGMENTS The authors thank Kevin L. Nunes, Michael C. Seto, Pamela M. Yates, and two anonymous reviewers for their very helpful comments on an earlier draft of this paper. This research was facilitated by a Social Sciences and Humanities Research Council of Canada Doctoral Fellowship and Ontario Graduate Scholarship awarded to the ﬁrst author. REFERENCES Abel GG. 1985. Use of pornography and erotica by sex offenders. Paper presented to the United States Attorney General’s Commission on Pornography, Houston, TX. Allen M, D’Alessio D, Brezgel K. 1995a. A meta-analysis summarizing the effects of pornography II: Aggression after exposure. Hum Commun Res 22:258–283. Allen M, Emmers T, Gebhardt L, Giery MA. 1995b. Exposure to pornography and acceptance of rape myths. J Commun 45:5–26. Boeringer S. 1994. Pornography and sexual aggression: Associations of violent and nonviolent depictions with rape and rape proclivity. Deviant Behav: An Interdisciplinary J 15:289–304. Bradford JM, Pawlak AE, Boulet JR. 1987. Bradford Sexual History Inventory (BSHI), Ottawa, Ontario, Canada: Royal Ottawa Hospital. Bradford JM, Pawlak AE, Boulet JR, Curry S. 1991. Bradford Sexual History Inventory (BSHI), Ottawa, Ontario, Canada: Royal Ottawa Hospital. Cohen J. 1988. Statistical Power Analysis for the Behavioral Sciences, 2nd edition, Hillsdale, NJ: Lawrence Erlbaum. Cohen J. 1992. A power primer. Psychol Bull 112:155–159. Condron MK, Nutter DE. 1988. A preliminary examination of the pornography experience of sex offenders, paraphiliacs, sexual dysfunction patients, and controls based on Meese Commission recommendations. J Sex Marital Ther 14:285–298. Cook RF, Fosen RH, Pacht A. 1971. Pornography and the sex offender: Pattern of previous exposure and arousal effects of pornographic stimuli. J Appl Psychol 55:503–511. Court JH. 1976. Pornography and sex crimes: A re-evaluation in the light of recent trends around the world. Int J Criminol Penology 5:129–157. De Vogel V, de Ruiter C, van Beek D, Mead G. 2004. Predictive validity of the SVR-20 and Static-99 in a Dutch sample of treated sex offenders. Law Hum Behav 28:235–251. Doren D. 2004. Stability of the interpretative risk percentages for the RRASOR and Static-99. Sex Abus: A J Res Treat 16:25–36. Firestone P, Bradford JM, Greenberg DM, McCoy M, Larose MR, Curry S. 1999. Prediction of recidivism in incest offenders. J Interpers Violence 14:511–531. Firestone P, Kingston DA, Wexler A, Bradford JM. 2006. Long term follow-up of exhibitionists: Psychological, phallometric and offence characteristics. J Am Acad Psychiatry Law 34: 349–359. Grubin D. 1998. Sex offending against children: Understanding the risk. Police Research Series Paper 99. London: Home Ofﬁce. Hanson RK. 1997. The development of a brief actuarial risk scale for sexual offence recidivism (User Report No. 1997–04). Ottawa: Department of the Solicitor General of Canada. Available from http://www.sgc.gc.ca/epub/corr/e199704/e199704.htm Hanson RK, Morton-Bourgon K. 2004. Predictors of Sexual Recidivism: An Updated Meta-Analysis, Ottawa: Public Safety and Emergency Preparedness Canada. Hanson RK, Thornton D. 1999. Static-99: Improving Actuarial Risk Assessment for Sex Offenders. Ottawa: Department of the Solicitor General of Canada. Hanson RK, Thornton D. 2000. Improving risk assessment for sex offenders: A comparison of three actuarial scales. Law Hum Behav 24:119–136. Harris AJR, Phenix A, Hanson RK, Thornton D. 2003. Static-99 Coding Rules: Revised 2003. Ottawa, ON: Department of the Solicitor General of Canada. Kimmel MS, Linders A. 1996. Does censorship make a difference? An aggregate empirical analysis of pornography and rape. J Psychol Hum Sex 8:1–20. Kingston DA, Firestone P, Moulden HM, Bradford JM. 2007. The utility of the diagnosis of pedophilia. A comparison of various classiﬁcation procedures. Arch Sex Behav 36:423–436. Kutchinsky B. 1991. Pornography and rape: Theory and practice. Int J Law Psychiatry 14:47–64. Lalumie` re ML, Harris GT, Quinsey VL, Rice ME. 2005. The Causes of Rape: Understanding Individual Differences in Male Propen- sity for Sexual Aggression. Washington, DC: American Psycho- logical Association. Malamuth NM. 1986. Predictors of naturalistic sexual aggression. J Pers Soc Psychol 50:953–962. Malamuth NM. 1998. The conﬂuence model as an organizing framework for research on sexually aggressive men: Risk moderators, imagined aggression, and pornography consump- tion. In: Geen RG, Donnerstein E (eds). Human Aggression: Theories, Research, and Implications for Social Policy, San Diego: CA: Academic Press, pp 229–245. Malamuth NM. 2003. Criminal and noncriminal sexual aggressors: Integrating psychopathy in a hierarchical-mediational conﬂuence model. In: Prentky RA, Janus ES, Seto MC (eds). Sexually Coercive Behavior: Understanding and Management, New York, NY: Annals of the New York Academy of Sciences, pp. 33–58. Malamuth NM, Check J. 1981. The effects of mass media exposure on acceptance of violence against women: a ﬁeld experiment. J Res Pers 15:436–446. 350 Kingston et al. Aggr. Behav. Malamuth NM, Huppin M. 2005. Pornography and teenagers: The importance of individual differences. Adolesc Med 16:315–326. Malamuth NM, Petipitan EV. 2007. The effects of pornography are moderated by men’s sexual aggression risk. In: Guinn DE (ed). Pornography: Driving the Demand for International Sex Trafﬁcking? Los Angeles: Captive Daughters Media, pp 125–143. Malamuth NM, Linz D, Heavey CL, Barnes G, Acker M. 1995. Using the conﬂuence model of sexual aggression to predict men’s conﬂict with women: a 10-year follow-up study. J Pers Soc Psychol 69:353–369. Malamuth NM, Addison T, Koss M. 2000. Pornography and sexual aggression: Are there reliable effects and can we understand them? Annu Rev Sex Res 11:26–91. Marshall WL. 1988. The use of sexually explicit stimuli by rapists, child molesters, and nonoffenders. J Sex Res 25:267–288. Marshall WL. 2000. Revisiting the use of pornography by sexual offenders: Implications for theory and practice. J Sex Aggression 6:67–77. Marshall WL, Seidman BT, Barbaree HE. 1991. The effects of prior exposure to erotic and nonerotic stimuli on the rape index. Ann Sex Res 4:209–220. Nugent PM, Kroner DG. 1996. Denial, response styles, and admittance of offenses among child molesters and rapists. J Interpers Violence 11:475–486. Nunes KL, Firestone P, Bradford JM, Greenberg DM, Broom I. 2002. A comparison of modiﬁed versions of the Static-99 and the Sex Offender Risk Appraisal Guide. Sex Abuse: A J Res Treat 14:253–269. Quinsey VL, Harris GT, Rice ME, Cormier CA. 1998. Violent offenders: Appraising and managing risk. Washington, DC: American Psychological Association. Rice ME, Quinsey VL, Harris GT. 1991. Sexual recidivism among child molesters released from a maximum security psychiatric institution. J Consult Clin Psychol 59:381–386. Rice ME, Harris GT, Lang C, Cormier C. 2006. Violent sex offences: How are they best measured from ofﬁcial records? Law Hum Behav 30:525–541. Roberts CF, Doren DM, Thornton D. 2002. Dimensions associated with assessments of sex offender recidivism risk. Crim Justice Behav 29:569–589. Seto MC, Eke AW. 2005. The criminal histories and later offending of child pornography offenders. Sex Abuse: A J Res Treat 17: 201–210. Seto MC, Eke AW. 2006. Extending the follow-up of child pornography offenders reported by Seto and Eke (2005). Poster presented at the 25th Annual Research and Treatment Con- ference for the Association for the Treatment of Sexual Abusers, Chicago, IL. Seto MC, Maric A, Barbaree HE. 2001. The role of pornography in the etiology of sexual aggression. Aggression Violent Behav 6:35–53. Seto MC, Cantor JM, Blanchard R. 2006. Child pornography offences are a valid diagnostic indicator of pedophilia. J Abnorm Psychol 115:610–615. Shim JW, Lee S, Paul B. 2007. Who responds to unsolicited sexually explicit materials on the internet? The role of individual differences. Cyberpsychol Behav 10:71–79. Vega V, Malamuth NM. 2007. Predicting sexual aggression: The role of pornography in the context of general and speciﬁc risk factors. Aggressive Behav 33:104–117. 351 Pornography Use and Sexual Aggression Aggr. Behav.