Rationale --------- In recent years, a deluge of personal disclosures of sexual victimization (i.e., #MeToo) and high-profile allegations of sexual misconduct, harassment, and assault have revitalized contemporary discussions of sexual aggression. As these stories make clear, sexual aggression can inflict a range of physical (i.e., physical injury, sexual / reproductive health consequences, death), mental (i.e., depression, anxiety, post-traumatic stress disorder, suicidality, etc.), and social harms (i.e., stigmatization, ostracism, honour killings, etc.) (Krug, Dahlberg, Mercy, Zwi, & Lozano, 2002). Although not a new issue, the current surge in these reports clearly reflects the high incidence and prevalence of sexual aggression; it has been estimated that more than 430,000 sexual assaults occurred in the United States in 2015 (Truman & Morgan, 2018), with the burden of victimization generally falling predominantly on women and girls (Snyder, 2000). Sexual aggression is not limited to the United States, and the World Health Organization has reported that in some countries, as many as one in four women may be sexually victimized (Krug et al., 2002). Sexual violence also occurs among adolescents. In fact, the U.S. Department of Justice has estimated that women between the ages of 16-19 are four times more likely to be victims of sexual aggression than the general public (Greenfeld, 1997). Acknowledging difficulties with integrating such findings, Bonino and colleagues (2006) have estimated that approximately 15% of adolescent women and 10% of adolescent men are victims of unwanted sexual attention or sexual violence. Importantly, adolescents may also be perpetrators of sexual aggression; a large national sample of American teens found that 5% of adolescents reported engaging in sexual aggression over a three year period (Ybarra, Mitchell, Hamburger, Diener-West, & Leaf, 2011). There have been longstanding concerns that pornography use contributes to sexual aggression (Commission on Obscenity and Pornography, 1971), and while conclusions continue to be contested (c.f. Ferguson & Hartley, 2009; Fisher, Kohut, Di Gioacchino, & Fedoroff, 2013; Kingston, Malamuth, Fedoroff, & Marshall, 2009), a growing body of evidence connects pornography use to self-reports of sexual aggression among adolescents. Cross-sectional and longitudinal research in the U.S., Norway, Italy, and Taiwan all indicate that adolescents who use more pornography report engaging in more sexual harassment and aggression (Bonino et al., 2006; Brown & L’Engle, 2009; Chang et al., 2016; Kennair & Bendixen, 2012; Ybarra et al., 2011). Indeed, recent meta-analytic work has found that the association between pornography use and sexual violence did not differ between adolescents and adults (Wright, Tokunaga, & Kraus, 2016). The Confluence Model of sexual aggression (Malamuth & Hald, 2017) offers a theoretical explanation for the role that pornography use plays in male sexual violence directed towards women. Of particular importance, it argues that not all male users are equally vulnerable to this influence. According to Confluence Model theory, sexual aggression is the product of the interactive confluence of proximate risk factors of sexual aggression, including particularly, the constructs of hostile masculinity and impersonal sexuality, and pornography use. Hostile masculinity has been described as a “narcissistic, insecure, defensive, hypersensitive, and hostile-distrustful orientation” (Malamuth & Hald, 2017, p. 54) towards women accompanied by the sexual desire to control and dominate women. Impersonal sexuality, in contrast, is said to be a “promiscuous detached orientation towards sexual relations” (Malamuth & Hald, 2017, p. 54), and is similar to Simpson and Gangestad’s (1991) conceptualization of sociosexuality. With respect to pornography use, the Confluence Model “contends that for a small subgroup of users, who already score high on other known risk factors of sexually aggression, pornography consumption may add ‘fuel to the fire’ and increase the risk of sexually aggressive attitudes and behaviors… by aiding the creation, reinforcement, or priming of (preexisting) sexually aggressive attitudes, cognitions, and emotions.” (Hald & Malamuth, 2015, p. 100). Although the role of pornography use is understudied compared to other aspects of the Confluence Model, available evidence supports the contention that pornography use can be a risk factor for sexual aggression. Specifically, cross-sectional research has indicated that pornography use is correlated with self-reported sexual aggression (Baer, Kohut, & Fisher, 2015; Malamuth, Addison, & Koss, 2000) and attitudes (Malamuth, Hald, & Koss, 2012) supportive of sexual aggression, but only among men who are predisposed to sexual aggression. Recent experimental work inspired by the Confluence Model has also found that exposure to pornography increased attitudes supporting violence towards women, but only among men who were low in agreeableness (Hald & Malamuth, 2015)—a distal risk factor that is thought to be mediated by hostile masculinity. Malamuth’s conceptualization of the Confluence Model is not without criticism. At present, the Confluence Model’s assertion that pornography plays a causal role in sexual aggression remains unverified. The available evidence concerning the interactive contributions of pornography use, hostile masculinity and impersonal sexuality comes exclusively from cross-sectional research. The sole experimental study did not examine interactions between hostile masculinity, impersonal sexuality and pornography exposure (Hald & Malamuth, 2015). It focused, instead, on more distal predictors of sexual aggression. For obvious ethical reasons, it also examined pornography-induced changes in attitudes toward sexual aggression than self-reported acts of sexual aggression. Although there are also a few older longitudinal studies that have tested some components of the Confluence Model (e.g., Malamuth, Linz, & Heavey, 1995), they predate the inclusion of pornography use in Confluence Model theory as one of the key risk factors for sexual aggression. Thus, longitudinal assessment of pornography’s role in the Confluence Model remains an imperative. The asserted causal relationship between pornography use and sexual aggression in the Confluence Model has also been challenged for failing to account for potential confounds (Baer et al., 2015). Sex drive is one such factor. It is independent of the construct of impersonal sexuality (Malamuth et al., 1995); it is correlated with pornography use (Baer et al., 2015); and it is implicated in paraphilic sexual interests (Dawson, Bannerman, & Lalumière, 2014), and self-reported sexual aggression (Malamuth et al., 1995). There are also connections—albeit, far from straightforward (van Anders, 2013)—between men’s intensity of sex drive and testosterone levels, and the regulation of sex drive through chemical / surgical castration is among the most effective treatments for sex offending recidivism (Kim, Benekos, & Merlo, 2016). Interestingly, men with higher levels of testosterone also report less nurturant content in their sexual fantasies (Goldey, Avery, & van Anders, 2014). Failing to control for constructs like sex drive and perhaps testosterone levels, in addition to other factors that potentially connect pornography use to sexual aggression (e.g., pubertal status, sensation seeking, impulsiveness, social desirability, etc.), risks inflating the magnitude of the direct relationship between these two constructs. The Confluence Model assertion that aggression inducing effects of pornography occur even when sexual violence is not depicted in the material has also been called into question (Baer, et al., 2015). Although Wright and colleagues (Wright et al., 2016) reported no significant differences in sexual aggression between the use of violent and non-violent pornographic materials in their meta-analysis, some studies of adolescents have reported substantial differences in the effect sizes of violent compared to non-violent pornography (e.g. Ybarra et al., 2011). In addition, Seto and Lalumière (2010) found that atypical sexual tendencies—including an interest in sexual violence among adolescents—was the single largest predictor of sexual compared to non-sexual offending. Perhaps most importantly, Baer and colleagues (2015) have found that the confluence of hostile masculinity and impersonal sexuality was related to the type of pornography men consume. Men who were high in both of these factors used violent pornography significantly more often than men who were low in these factors. In other words, men who are high in risk of sexual aggression as defined by the Confluence Model appeared to consume fundamentally different types of pornography then men who are at low risk of sexual aggression. Current Study ------------- To provide the first longitudinal assessment of Confluence Model theory, the current study seeks to test the association between pornography use and sexual aggression using two independent longitudinal panel samples of male Croatian adolescents. Following the conceptual model, of primary interest is whether antecedent levels of hostile masculinity, impersonal sexuality, and pornography use are associated with subsequent changes in self-reported sexual aggression. If pornography use can be said to cause sexual aggression for men who are at risk of engaging in sexual aggression, as stipulated by the Confluence Model (Hald & Malamuth, 2015; Malamuth et al., 2000; Malamuth & Hald, 2017), two- and three-way interactions between pornography use and the risk factors of hostile masculinity and impersonal sexuality should be associated with increases in the probability of sexual aggression over time. The data required for such an analysis also permit the examination of contemporaneous associations between sexual aggression and the Confluence of hostile masculinity, impersonal sexuality, and pornography use across repeated measurement occasions. In effect, such an approach is akin to aggregating the results of a series of cross-sectional analyses and will be useful for connecting the current data to past research findings. Of further interest is whether any relationships that may emerge between pornography use and sexual aggression will remain significant after controlling for relevant confounds (e.g., masturbation frequency as a proxy for sex drive, testosterone levels, impulsiveness, sensation seeking, and social desirability). Finally, we also wish to determine if men who are high in both hostile masculinity and impersonal sexuality are more likely to consume aggressive/violent pornography. Given the lack of longitudinal explorations into the central theoretical conceptualization of the link between pornography use and sexual aggression, this study has the potential to advance our understanding of the target relationship in several ways. First, to the best of our knowledge, this study is the first attempt to examine the Confluence Model among adolescents. Second, it will test the causal assertions concerning pornography use and sexual aggression more precisely and robustly than past research. Finally, it will provide empirical insights that can be used to gauge the recent claims that sex drive may account for the associations between pornography use and sexual aggression (Baer et al., 2015).