تمایلات جنسی اجتماعی به عنوان پیش بینی کننده آزار و اذیت جنسی و اجبار در دانش آموزان دختر و پسر

Abstract Sexual harassment and coercion have mainly been considered from a sex difference perspective. While traditional social science theories have explained harassment as male dominance of females, the evolutionary perspective has suggested that sex differences in the desire for sex are a better explanation. This study attempts to address individual differences associated with harassment from an evolutionary perspective. Considering previous research that has found links between sociosexual orientation inventory (SOI) and harassment, we consider whether this association can be replicated in a large, representative sample of high school students (N=1199) from a highly egalitarian culture. Expanding the previous studies which mainly focused on male perpetrators and female victims, we also examine females and males as both perpetrators and as victims. We believe that unrestricted sociosexuality motivates people to test whether others are interested in short-term sexual relations in ways that sometimes might be defined as harassment. Furthermore, unrestricted individuals signal their sociosexual orientation, and while they do not desire all individuals that react to these signals with sexual advances, they attract much more sexual advances than individuals with restricted sociosexual orientations, especially from other unrestricted members of the opposite sex. This more or less unconscious signaling thus makes them exploitable, i.e., harassable. We find that SOI is a predictor for sexual harassment and coercion among high school students. The paper concludes that, as expected, unrestricted sociosexuality predicts being both a perpetrator and a victim of both same-sex and opposite-sex harassment.

1. Introduction Sexual harassment at work represents a considerable problem in the management of organizations (UN, 2006). Early studies portrayed the traditional view of male perpetrators and adverse consequences for female victims (Fitzgerald, 1993). The feminist perspective has focused on men's need for power as the root cause of conflict between the sexes, seeing sexual harassment as a tool of patriarchy (Browne, 2006 and Studd, 1996). Sexual harassment is linked to the need to dominate women accompanied by hostile attitudes (including condoning forced sex) and men's use of their organizational power to oppress their female subordinates (Smith and Konik, 2011 and Tangri et al., 1982). Evolutionary psychology (EP) presents an alternative perspective claiming that sex is the driving force in sexual harassment—not males' need for power over women (Buss, 1996 and Vandermassen, 2011). We support Vandermassen's (2011) perspective that feminism and EP may mutually inform each other, but one needs to treat both approaches as testable scientific theories (Buss & Schmitt, 2011). As such, the contemporary discussion in the journal Sex Roles ( Smith & Konik, 2011) may reflect a watershed in the debate between these two perspectives, resulting in less polemic and more empirical testing (although see Liesen, 2011). Sexual harassment is defined as unwanted sexually oriented behavior such as offensive sexual attention or hostile behaviors that focus on gender (Fitzgerald et al., 2001 and Studd, 1996). It is worth noting that “unwanted” and “offensive” are from the perspective of the person experiencing the harassment. Similar acts from more attractive persons might not be considered harassment, but rather be perceived as flirtatious (Browne, 2006). Sexual harassment covers various acts ranging from sexual comments (obscene language and jokes) and spreading rumors, through inappropriate sexual advances to showing pictures of nudity and sex. Sexual coercion is the use of physical force to obtain sex. Most research on sexual harassment, conducted within the hierarchical organizational context, has studied adult male perpetrators and female targets. However, sexual harassment is common among high school girls and also among boys (American Association of University Women, 2001, Bendixen and Kennair, 2008 and Witkowska, 2005). Studies report 12-month prevalence rates in the 40%–50% range for sexist jokes, degrading and obscene sexual comments, and homophobic insults. Nearly one in four high school students reported the spreading of sexual rumors, being shown pictures of nudity, and being asked for sexual favors. Apart from homophobic insults, more girls than boys reported harassment. Coercion is much less common but is still reported by a significant proportion of students ( Bendixen & Kennair, 2008). Several findings support the sexual motives hypothesis advanced by EP (Studd, 1996). The evidence is particularly strong that victims of sexual harassment are not random women. They are disproportionally young, single, and employed as low-level office workers, waitresses, or cooks in the service sector (Browne, 2006 and Studd, 1996). From a male perspective, gaining sexual access to a young and single woman would, potentially, bring greater reproductive benefits than sexual access to an older married woman. Reliable patterns of sexual harassment and coercion may be predicted by EP (Buss & Kenrick, 1998). Over evolutionary time, the use of coercive means to gain sexual access to unconsenting women increased the reproductive fitness for males but not for females. Differences in evolved sexual strategies (Buss and Schmitt, 1993 and Trivers, 1972) and evolved mental mechanisms between the sexes explain why women perceive more situations as sexual harassment or why more men are perpetrators of sexual harassment (Studd & Gattiker, 1991), why more males overperceive sexual intent from the opposite sex (Haselton, 2003), or why the same behavior by different men is perceived differently by women (Sheets & Braver, 1999). Evolved sex differences have been far more subject to investigation than individual differences, while there are greater behavioral differences within the same sex than between the sexes (Simpson & Gangestad, 1991). Currently, there is growing interest in evolved individual differences and explanations of within-sex differences (Buss & Hawley, 2010). This paper addresses whether there are evolved within-sex differences that predict sexual harassment and coercion of others and predictors of being harassed or coerced. Early findings (Bendixen & Kennair, 2008) suggest that there is a large overlap between those who harass others and those who are targets of harassment. The etiology of this is not understood, and it remains open to question whether harassing others increases the risk of being harassed (or vice versa) or whether harassing others and being a target of harassment share common individually differing antecedents. Who is harassed, and how are they identified as exploitable by harassers? Buss and Duntley (2008) suggest that there are different types of exploitation—and there will be different markers that suggest whether or not one is prone to being exploited in a specific area. Such areas of vulnerability are, therefore, person and context specific. Evolved mental mechanisms will be specifically orientated toward perceiving such signs of exploitability. For example, Sakaguchi and Hasegawa (2006) found that Japanese males used awkward gait as a cue to vulnerability for inappropriate touching. Moreover, the male subjects' judgments were able to predict real-life sexual harassment, as reported by the female walkers. Exploitability is thus a specific quality of an individual that conveys to potential exploiters the following information (Buss & Duntley, 2008). Sakaguchi and Hasegawa (2007) found that unexpected sexual attention was predicted by the targets' unrestricted sociosexuality as measured by Simpson and Gangestad's (1991) sociosexual orientation inventory (SOI; a stable individual behavioral and attitudinal characteristic that predicts sexual behavior across many different situations). Inappropriate touching of body parts was, however, not predicted by SOI. Hence, high-SOI persons are more subject to harassment, but not necessarily to coercion in this study. Individual differences in SOI are discernible to observers: SOI may be estimated through visual cues such as facial features and body language (Boothroyd et al., 2008, Gangestad et al., 1992 and Sakaguchi and Hasegawa, 2007), and high-SOI females are rated to be more attractive than low-SOI females. In the light of Sexual Strategies Theory (Buss & Schmitt, 1993), the major sex difference in SOI (Schmitt, 2005) is due to males being more open to short-term relations (Clark and Hatfield, 1989 and Kennair et al., 2009), while females consider sexual relations more from a long-term perspective (Campbell, 2008, Haselton and Buss, 2001 and Surbey and Conohan, 2000). Individuals with high SOI scores thus seem to signal their greater interest in short-term sexual affairs. These signals are perceived by others (Sakaguchi & Hasegawa, 2007). Compared to individuals with a restricted sociosexuality, they will, therefore, attract more sexual attention—not least from other unrestricted individuals of the opposite sex. If we compared two females—one with an unrestricted sociosexuality and one with lower SOI scores—in a social setting, we might find the following: unrestricted sociosexuality elicits much more sexual solicitations. If the numbers of solicitations were similar, the female with the lower SOI scores would find more of these harassing, as fewer of the attempts to interest her in having casual sex would be desirable. But the total number of advances is expected to be significantly higher due to the female with higher SOI scores signaling availability, and not all of these are welcome; unrestricted females will not find all males attractive. This may add up to a higher total number of experienced harassing advances. In addition, intrasexual competition would put high-SOI subjects at a higher risk of being subject to insults, slander, rumors, and other forms of derogation ( Schmitt & Buss, 1996). High-SOI subjects are expected to report more sexual harassment because they are the object of both unwanted sexual advances from members of the opposite sex as well as derogation from same-sex competitors. We suggest that the perpetrators might perceive their victim's unrestricted sociosexuality as a cue to exploitability; high-SOI subjects will signal harassability. High-SOI individuals are probably not aware of the effects of advertising their unrestricted sociosexuality to both desirable and undesirable potential sexual partners. It is the increased amount of undesirable partners that increases the high-SOI individuals' experience of harassability. While experiencing more harassment, signaling high SOI also increases the number of desirable sexual encounters. Haselton (2003) found that unrestricted sociosexuality was positively associated with the total number of misperceptions of sexual intention. Misperception and biased communication between the sexes might increase the level of perceived sexual harassment in society. Yost and Zurbriggen (2006) found that SOI predicted coercion. In their limited sample of adults, they also noted that females with high SOI scores had more fantasies about sexually dominating others. We posit that behavior described as sexual harassment is primarily behavior that intends to investigate whether the initial perception of an unrestricted sociosexuality is correct, and whether “hooking up” or other short-term sexual relations are possible. Thus, sexually laden communication that signals interest in physical aspects of sex and sexual surgency, rather than love and commitment, is the hallmark of these interactions. Individuals with a restricted sociosexuality will receive much less undesirable attention—due to not advertising an interest in casual sex and signaling exploitability (behavior that increases the number of undesired sexual advances) to the same degree. If the intention is to dominate or suppress, a perpetrator of sexual harassment ought to target individuals with restricted sociosexuality—as those individuals would be most upset by the behavior. We believe that the main intention is to solicit sex, and thus targeting individuals that advertise their unrestricted sociosexuality is the most effective strategy. Consequently, higher SOI scores will predict an increased likelihood of being harassed, despite a greater interest in receiving sexual attention from attractive partners. Furthermore, sexually aggressive and harassing behavior, understood as attempts to procure casual sex rather than a romantic relationship, is also predicted by increased SOI scores. In this study, the association between SOI and sexual harassment and coercion is examined using a large sample of gender egalitarian high school students in the context of more traditional social science variables. Exposure to pornography, coercive sexual attitudes, and sexist attitudes are all consistently found to correlate with self-reported coercion (Begany and Milburn, 2002, Glick and Fiske, 1996, Malamuth et al., 2000 and Yost and Zurbriggen, 2006). For this reason, measures of porn exposure (type and frequency) as well as attitudes toward coercion (rape myths) and blatant, hostile sexism are included in our analysis. In addition, the analyses are extended to measures of being sexually harassed and coerced, as the prior literature on this particular issue is scarce.

. Results 3.1. Descriptives Males reported markedly more lenient attitudes toward unrestricted sex than females (Table 1), which held true across all attitude items. Small sex differences were evident for the behavior variables measuring the number of different sex partners reported in the past year. This was due to a larger proportion of females (63.4%) than males (47.7%) reporting having at least one sex partner during the past year. The average number of different partners in the past year among those who reported having had sexual partners was slightly higher for males (M=2.6) than for females (M=2.1). Similar effects were evident for the variable measuring the number of different one-night stands. The proportion of one-night stands was slightly higher for females than for males (45.9% vs. 40.8%), but the average number of one-night stands was similar among those who reported one or more one-night stands (males M=2.2 and females M=2.1). Table 1. Mean variable and scale scores and S.D. for females (n=566–600) and males (n=530–595) with gender effect sizes (Cohen's d and h) Scales Females Males t/χ2 ES Mean S.D. Mean S.D. 1. SOI components a. N of different partners (0–30) 1.34 2.01 1.22 2.13 0.90 −.07 b. N of different one-night stands (0–15) 0.98 1.58 0.88 1.57 1.11 −.05 c. SOI-Attitudes (1–10) 4.25 2.22 6.07 2.42 −13.53⁎⁎⁎ .79 2. Exposure to pornography a. Erotic magazines (0, 1) 0.17 0.37 0.29 0.46 a 25.45⁎⁎⁎ b.29 b. Soft-core (X-rated) (0, 1) 0.34 0.47 0.79 0.41 a 243.12⁎⁎⁎ b.94 c. Hard-core (XXX-rated) (0, 1) 0.16 0.37 0.73 0.45 a 380.28⁎⁎⁎ b1.23 d. Frequency exposure (1, 5) 1.59 0.70 3.31 1.17 −30.84⁎⁎⁎ 1.79 3. Coercive Sexual Attitudes (1–5) 2.08 0.76 2.61 0.75 −11.92⁎⁎⁎ .69 4. Classical Sexism (1–5) 1.99 0.47 2.68 0.61 −22.02⁎⁎⁎ 1.27 5. Sexually Harassed (0–1) 0.30 0.25 0.27 0.24 1.91† −.11 6. Sexually Coerced (0, 1) 0.42 0.49 0.26 0.44 a 33.81⁎⁎⁎ b−.34 7. Sexually Harassing (0–1) 0.12 0.17 0.21 0.22 −7.33⁎⁎⁎ .42 8. Sexually Coercive (0, 1) 0.03 0.17 0.10 0.30 a 26.05⁎⁎⁎ b.30 Numbers in parenthesis are range of scores (continuous [–] and category [,]). a χ2 for 2×2 tables. b Effect sizes (h) are for proportions. † p<.10. ⁎⁎⁎ p<.001. Table options A much higher proportion of males reported consuming erotic magazines as well as soft-core and hard-core pornography. In particular, the frequency of exposure to erotica and pornographic magazines and movies was markedly higher for males than for females (d=1.79). Male respondents also reported more lenient attitudes toward coercive sex (M=2.61) compared to female respondents (M=2.08), producing a moderately high effect size. Females reported being slightly more sexually harassed than males, and far more females than males reported being coerced (42% of females vs. 26% of males) (Table 1). The effect sizes for these sex differences were small to moderate. A moderate sex effect was found for harassing and coercing others as male respondents reported stronger involvement in both harassing others and the use of coercion (10% of males, 3% of females). 3.2. Correlations The two SOI components were significantly associated with Porn Exposure for both sexes (r's ranging from .11 to .25). The SOI components were insignificantly associated with Coercive Attitudes and Classical Sexism in females. In males, the behavioral component of SOI was marginally associated with Coercive Attitudes (r=.08), but both SOI components were significantly associated with Classical Sexism. Furthermore, Porn Exposure was not significantly associated with Coercive Attitudes in either sex and only weakly associated with Classical Sexism in males (r=.09). Coercive Attitudes and Sexism were moderately associated in males and females (r=.30). The outcome variables Being Harassed and Harassing Others were moderately associated with the SOI components and Porn Exposure in females and males. The corresponding associations with Coercive Attitudes and Classical Sexism were nonsignificant in females, and between r=.09 and r=.23 in males ( Table 2). Table 2. Zero-order correlations (Pearson r) among the scales 2 3 4 5 6 7 8 9 Scales Females (n=542–600) 1. SOI-Behavior (2) .39⁎⁎⁎ .24⁎⁎⁎ .02 .03 .31⁎⁎⁎ .26⁎⁎⁎ .18⁎⁎⁎ .07 2. SOI-Attitudes (3) .19⁎⁎⁎ −.06 −.02 .24⁎⁎⁎ .13⁎⁎ .17⁎⁎⁎ .04 3. Porn Exposure (4) .03 .03 .30⁎⁎⁎ .22⁎⁎⁎ .29⁎⁎⁎ .05 4. Coercive Attitudes (9) .30⁎⁎⁎ .02 .01 .05 .10⁎ 5. Classical Sexism (7) .02 .04 .00 .07 6. Sexually Harassed (9) .54⁎⁎⁎ .63⁎⁎⁎ .22⁎⁎⁎ 7. Sexually Coerced (4) .34⁎⁎⁎ .18⁎⁎⁎ 8. Sexually Harassing (9) .28⁎⁎⁎ 9. Sexually Coercive (4) Males (n=511–593) 1. SOI-Behavior (2) .32⁎⁎⁎ .11⁎ .08† .10⁎ .29⁎⁎⁎ .11⁎ .22⁎⁎⁎ .06 2. SOI-Attitudes (3) .25⁎⁎⁎ .04 .15⁎⁎⁎ .21⁎⁎⁎ .11⁎ .30⁎⁎⁎ .09⁎ 3. Porn Exposure (4) .06 .09⁎ .23⁎⁎ .12⁎⁎ .28⁎⁎⁎ .12⁎⁎ 4. Coercive Attitudes (9) .30⁎⁎⁎ .12⁎⁎ .09⁎ .16⁎⁎ .13⁎⁎ 5. Classical Sexism (7) .18⁎⁎⁎ .17⁎⁎⁎ .23⁎⁎⁎ .11⁎⁎ 6. Sexually Harassed (9) .42⁎⁎⁎ .65⁎⁎⁎ .26⁎⁎⁎ 7. Sexually Coerced (4) .38⁎⁎⁎ .34⁎⁎⁎ 8. Sexually Harassing (9) .39⁎⁎⁎ 9. Sexually Coercive (4) Numbers in parenthesis are number of scale items. † p<.10. ⁎ p<.05. ⁎⁎ p<.01. ⁎⁎⁎ p<.001. Table options Being coerced was moderately associated with the SOI components and Porn Exposure in females. These associations were significant but weaker in males. However, both Coercive Attitudes and Sexism correlated significantly with being coerced in males. Being coercive was insignificantly associated with the SOI components, Porn Exposure, and Sexism in females and only weakly associated with Coercive Attitudes. Corresponding associations for males were generally weak but significant (nonsignificant only for SOI-Behavior). The two components of SOI were moderately associated, but the disattenuated correlations (assuming no measurement errors) were strong (.56 for females and .47 for males). Despite the lower number of scale items, these associations were highly comparable to those reported by Webster and Bryan (2007) for latent constructs. Finally, we found that being harassed was substantially associated with harassing others in the male and female samples (r's above .60). This finding indicates that there was a large overlap between targets and perpetrators of sexual harassment in these samples. Although the correlations were lower, this also held for dichotomously coded sexual coercion scales. Adding to the complexity of this phenomenon, males and females who reported being exposed to high levels of sexual harassment also tended to report being coerced (r=.54 for females and r=.42 for males). Neither the student's SES nor their age cohort showed any association with the sexual harassment or coercion scales, although the latter correlated weakly with the SOI components. 3.3. Predictors of being sexually harassed A hierarchical multiple regression was performed between being sexually harassed and SOI-Behavior and SOI-Attitudes in Block 1 (Model 1), Porn Exposure in Block 2 (Model 2), and Coercive Attitudes and Classical Sexism in Block 3 (Model 3). This allowed for the study of mediator effects and direct tests of Prediction 3. Interaction effects and the effect of age-cohort over and beyond that of the SOI components were entered in Block 4. Table 3 displays the unstandardized regression coefficients (B), standard error of the B, standardized regression coefficients (β), the corresponding t values, and the level of significance for each predictor at each step of the analysis. Having an unrestricted sociosexual orientation markedly increased the risk of being harassed in females. The effect was stronger for the behavioral component of the SOI than for the attitudinal component. As can be seen from Table 3, the SOI components remained significant even after controlling for the contribution of the other variables (but the effect of SOI-Attitudes was only marginally significant in Model 2 and Model 3, p<.10). As seen in the lower panel of Table 3, the two SOI components accounted for fully 11.2% of the variance in being harassed in Model 1 (adjusted R2 was 10.9%). Porn Exposure significantly predicted being harassed, adding 5.2% to the explained variance over and above that of SOI. Porn Exposure also mediated some of the effect of SOI-Attitudes. As would be expected from the zero-order correlations, Coercive Attitudes and Classical Sexism did not contribute to the prediction. None of the interactions significantly predicted being harassed, and age-cohort did not contribute to the explained variance over and above that of SOI and Porn Exposure. Table 3. Predictors of being sexually harassed Predictors Model 1 Model 2 Model 3 B S.E. B β t B S.E. B β t B S.E. B β t Block 1 SOI-Behavior .077 .012 .278 6.35⁎⁎ .064 .012 .234 5.41⁎⁎ .065 .012 .235 5.41⁎⁎ SOI-Attitudes .012 .005 .106 2.41⁎ .009 .005 .079 1.85† .009 .005 .078 1.80† Block 2 Porn Exposure .078 .013 .235 5.85⁎⁎ .078 .013 .236 5.85⁎⁎ Block 3 Coercive Sexual Att. −.005 .013 −.016 −0.39 Classical sexism .003 .022 .006 0.15 F change 35.05⁎⁎ 34.22⁎⁎ .08 R² change .112 .052 .000 R² adjusted .109 .160 .157 Females (n=555). † p<.10. ⁎ p<.05. ⁎⁎ p<.01. Table options Similar to the results in the female sample, males with an unrestricted sociosexual orientation were more likely to be sexually harassed (Table 4). Again, the behavioral component was a far better predictor for being sexually harassed than the attitudinal component. The two SOI components accounted for 9.3% of the variance in being harassed. Adding Porn Exposure at Step 2 significantly increased the explained variance, but the effect of the SOI variables remained significant (SOI-Attitudes only marginally, p<.10). Entering Coercive Attitudes and Classical Sexism at Step 3 contributed significantly to the explained variance of being sexually harassed for males (1.4%), but only the latter reached significance. None of the interactions (or age-cohort) contributed to the explained variance in being sexually harassed. Table 4. Predictors of being sexually harassed Predictors Model 1 Model 2 Model 3 B S.E. B β t B S.E. B β t B S.E. B β t Block 1 SOI-Behavior .061 .011 .237 5.37⁎⁎ .060 .011 .231 5.31⁎⁎ .058 .011 .224 5.17⁎⁎ SOI-Attitudes .013 .004 .129 2.92⁎⁎ .009 .004 .088 1.97⁎ .007 .004 .074 1.64† Block 2 Porn Exposure .061 .015 .176 4.14⁎⁎ .058 .015 .168 3.98⁎⁎ Block 3 Coercive Sexual Att. .013 .014 .039 0.91 Classical Sexism .041 .017 .104 2.40⁎ F change 26.47⁎⁎ 17.13⁎⁎ 4.30⁎ R2 change .093 .029 .014 R2 adjusted .089 .117 .128 Males (n=520). † p<.10. ⁎ p<.05. ⁎⁎ p<.01. Table options 3.3.1. Predictors of being sexually harassed by same-sex or opposite-sex peers Descriptive analysis on the reduced four-item harassed scale showed that same-sex harassment was more prevalent than opposite-sex harassment for males. For females, opposite-sex harassment was more prevalent. This is consistent with the finding that males reported stronger involvement in sexual harassment of others (Table 1). The overlap for being harassed by a male peer as well as by a female peer was considerable (r=.71 for both sexes). To study the effect of the four predictors of same-sex and opposite-sex harassment in male and female respondents, we carried out four separate multiple regressions ( Table 5). Table 5. Predictors of same-sex and opposite-sex sexual harassment by peers (reduced four-item scale) Same sex Opposite sex β t β t Females SOI-Behavior .137 3.00⁎⁎ .123 2.71⁎⁎ SOI-Attitudes −.003 −.10 .062 1.38 Porn Exposure .138 3.20⁎⁎ .190 4.50⁎⁎ Coercive Sexual Attitudes −.007 −.16 −.070 −1.62 Classical Sexism −.063 −1.44 −.043 −1.01 Males SOI-Behavior .125 2.81⁎⁎ .217 4.97⁎⁎ SOI-Attitudes .061 1.33 .102 2.27⁎ Porn Exposure .177 4.06⁎⁎ .148 3.49⁎⁎ Coercive Sexual Attitudes .005 .10 .052 1.21 Classical Sexism .058 1.29 .078 1.78† † p<.10. ⁎ p<.05. ⁎⁎ p<.01. Table options The behavior component of SOI turned out to be a consistent predictor for same-sex and opposite-sex harassment in both sexes. This was also true for Porn Exposure. The attitudes component of SOI was significant only for opposite-sex harassment in male respondents when the other predictors were included in the model. Coercive Sexual Attitudes failed to predict same-sex or opposite-sex harassment, and Classical Sexism was a marginal predictor only for opposite-sex harassment in males (p<.10). 3.4. Predictors of sexually harassing others For female respondents, the attitudinal and behavioral components of SOI were significantly associated with harassing others. The effect of the two SOI predictors was equal, suggesting that higher scores on SOI-Behavior or SOI-Attitudes were related to stronger involvement in harassing others. The SOI components accounted for 4.3% of the variance in the outcome variable. Porn Exposure was positively associated with harassing others and added 6.4% to the explained variance. Porn Exposure mediated some of the effects of the SOI components, but SOI still predicted harassing others in the final model. Coercive Attitudes and Classical Sexism was not related to the outcome variable (Model 3). None of the interactions turned out to be significantly related to harassment (or age-cohort) and hence are not reported here (Table 6). Table 6. Predictors of sexually harassing others Predictors Model 1 Model 2 Model 3 B S.E. B β t B S.E. B β t B S.E. B β t Block 1 SOI-Behavior .026 .009 .134 2.96⁎⁎ .016 .008 .085 1.90† .016 .009 .083 1.86† SOI-Attitudes .009 .004 .114 2.52⁎ .007 .003 .085 1.92† .007 .003 .088 1.99⁎ Block 2 Porn Exposure .060 .010 .261 6.30⁎⁎ .060 .010 .260 6.26⁎⁎ Block 3 Coercive Sexual Att. .009 .010 .040 0.95 Classical Sexism −.007 .016 −.019 −0.45 F change 12.62⁎⁎ 39.65⁎⁎ .47 R2 change .043 .064 .002 R2 adjusted .040 .102 .101 Females (n=558). † p<.10. ⁎ p<.05. ⁎⁎ p<.01. Table options Similar to the female sample, males' scores on the SOI components predicted sexual harassment of others, explaining 9.6% of the variance. As can be seen in Table 7, Porn Exposure added 4.3% to the explained variance. Adding Porn Exposure to the equation did, however, not influence the effect of the two SOI components. Throughout the three models, SOI-Attitudes was the stronger predictor of the two SOI components. Adding Coercive Attitudes and Classical Sexism in Block 3 gave a significant increase in the explained variance (3.3%) over and above that of SOI and Porn Exposure, but the effect of the prior predictors remained unaffected. None of the interactions or age cohort was significantly associated with harassing others in male respondents. Table 7. Predictors of sexually harassing others Predictors Model 1 Model 2 Model 3 B S.E. B β t B S.E. B β t B S.E. B β t Block 1 SOI-Behavior .031 .010 .135 3.09⁎⁎ .030 .010 .129 3.01⁎⁎ .027 .010 .117 2.79⁎⁎ SOI-Attitudes .021 .004 .238 5.45⁎⁎ .017 .004 .187 4.26⁎⁎ .015 .004 .167 3.85⁎⁎ Block 2 Porn Exposure .066 .013 .214 5.10⁎⁎ .063 .013 .201 4.89⁎⁎ Block 3 Coercive Sexual Att. .023 .012 .079 1.89† Classical Sexism .051 .015 .145 3.42⁎⁎ F change 27.72⁎⁎ 26.02⁎⁎ 10.39⁎⁎ R² change .096 .043 .033 R² adjusted .092 .133 .163 Males (n=527). † p<.10. ⁎⁎ p<.01. Table options 3.5. Predictors of being coerced A direct logistic regression analysis was performed on being coerced (no/yes) as outcome and the two SOI components, Porn Exposure, Coercive Attitudes, and Classical Sexism as predictors. Through a simple Wald test, an evaluation of individual predictors' contribution to a model is made available. The test is considered conservative (increased type II error) (Tabachnick & Fidell, 2001). The logistic regression for female respondents (n=557) showed that high scores on SOI-Behavior, but not SOI-Attitudes, significantly increased the odds of being coerced by 70% (Wald [χ2]=22.11, df=1, p<.001, OR=1.7). Porn Exposure increased the risk of being coerced by 1.6 (Wald=14.03, df=1, p<.001). Coercive Attitudes and Classical Sexism did not predict being coerced in female respondents. The effect of interactions and age-cohort was nonsignificant. The predictors accounted for 12.3% (Nagelkerke R2) of the variance in being coerced, correctly identifying 42.8% of the observed coerced cases. For male respondents (n=527), the odds of being coerced was marginally significantly associated with Porn Exposure (Wald=3.34, df=1, p<.10, OR=1.3). The SOI components did not predict being coerced, nor did Coercive Attitudes. The best predictor for being coerced was Classical Sexism (Wald=6.82, df=1, p<.01, OR=1.6)—none of the interactions (nor age-cohort) were significantly associated to being coerced. The predictors accounted for only 5.7% of the variance, correctly identifying only 4 of the 134 cases of men that reported being coerced. 3.5.1. Predictors of being coerced by same-sex or opposite-sex peers (one item) Female respondents reported being sexually squeezed far more by male peers than by female peers (18.6% and 2%, respectively). The corresponding figures for males were 5.6% (same-sex peer) and 7.3% (opposite-sex peer). Evidently, not only males are involved in this form of coercion, but coercion is essentially a male toward female phenomenon. For this reason, the analysis discussed below is restricted to females (n=555) reporting on unwanted sexual squeezing from male peers. None of the other logistic regression analyses (female on female, male on male, female on male) produced any meaningful or reliable results. The results closely mirrored those reported above for females being coerced. While SOI-Behavior significantly increased the odds of being coerced by a man by 1.4 (Wald=6.06, df=1, p<.05), SOI-Attitudes was not related to the odds of being coerced by a man over and above the other predictors. Porn Exposure marginally increased the odds of being coerced (Wald=2.99, df=1, p<.10, OR=1.3), while Coercive Attitudes and Classical Sexism failed to predict being coerced by a man. The overall explained variance accounted for by the predictors was low, only 4.1%. 3.6. Predictors of coercing others (males only) Since only 3% of the female sample reported being physically coercive, meaningful predictions could only be performed for males. Applying a hierarchic logistic regression analysis to prediction of being coercive (regardless of sex of target) showed that neither of the two SOI components predicted coercion in male respondents. Only Porn Exposure (Wald=4.38, df=1, p<.05, OR=1.7) and Coercive Attitudes predicted coercion (Wald=4.31, df=1, p<.05, OR=1.6). None of the interactions were significant. Overall, the predictors accounted for merely 6.4% of the variance in being physically coercive, but none of the 54 observed cases of coercion were correctly identified.