Initiation, Desistance, and Persistence of Men's Sexual Coercion.

The investigation of the course of sexually coercive behavior informs treatment and prevention. Those who engage in transitory sexually coercive behavior may be more amenable to treatment than those whose sexual coercion is persistent. Moreover, different courses of sexual coercion may also provide clues for the optimal timing of prevention efforts. The purpose of the current study was to examine the course of sexual coercion via self-report and laboratory measures in a large culturally and geographically diverse sample of college men.

The risk for sexual reoffense among perpetrators of sexual coercion is commonly perceived as high and persistent. Indeed, there is evidence of sexual offenders being at higher risk for reoffense than other types of offenders and remaining at risk for reoffense for periods of 25 years. Nevertheless, across studies, the recidivism rates among sexual offenders are relatively low. In a review of 61 studies and nearly 24,000 sexual offenders, the recidivism rate for new sexual offenses within a 4- to 5-year period was 13.4%. However, recidivism rates are most commonly based on official arrest data, which may underestimate actual rates of offending.

Some subtypes of sexual coercers may be at greater risk for recidivism than others. Hall and Hirschman (1991) proposed a theoretical model in which some sexual coercers engage in sexual coercion as a function of developmentally related personality problems, whereas other sexual coercers are motivated by situational and state-dependent physiological, cognitive, and affective factors. The most common type of sexual coercer was posited to be primarily motivated by cognitive distortions. These men may not appear deviant in terms of personality characteristics or sexual arousal and are commonly represented among acquaintance sexual coercers. Sexual coercers with developmentally related personality problems have more pervasive antisocial characteristics than do other sexual coercers. Antisocial sexual coercers and less pathological sexual coercers have consistently been identified by other researchers. The antisocial group has been found to be at greater risk for reoffense than other groups, even following treatment interventions.

In her theory of general antisocial behavior, Moffitt (1993) proposed a chronic, life-course-persistent trajectory as well as a second trajectory that is adolescence limited. Similar to the Hall and Hirschman (1991) model, chronic offenders were described as being influenced by developmental factors, such as temperament and family, whereas adolescence-limited offenders were described as being more influenced by situational factors, such as peers. Across community and high-risk samples of boys followed from school entry to early adolescence, distinct stable and declining patterns of antisocial behavior have been identified. In a longitudinal study of a general population at age 26, 11% of chronic offenders had been convicted for violence against women, including rape, versus only 1% of adolescence-limited offenders. Chronic offenders constitute a small group, particularly in community samples.

Men who first initiate sexually coercive behavior in adulthood have also been identified. Abbey and McAuslan (2004) conducted a 1-year longitudinal study of the initiation, desistance, and persistence of sexual coercion among 197 college men. Five percent of the men did not initially report sexual coercion but reported sexual coercion at the 1-year follow-up. Twenty-six percent of the men initially reported sexual coercion but desisted. Nine percent of the men reported sexual coercion at both assessments. Consistent with conceptualizations of chronic sexual coercers, men who persistently engaged in sexual coercion exhibited more risk factors for sexual coercion than did men who initiated, desisted, or did not engage in sexual coercion. Men who reported being sexually coercive at both assessments exhibited more hostile attitudes toward women, more sexual experiences, and more adolescent delinquency than did men who were not sexually coercive or than men who initially reported being sexually coercive and then desisted.

Hostile attitudes toward women, sexual experiences, and adolescent delinquency are risk factors associated with sexual coercion from Malamuth and colleagues' confluence model. The confluence model includes two paths, hostile masculinity and impersonal sex, that are associated with men's sexually coercive behavior. The hostile masculinity path is hierarchical, consisting of variables that are increasingly specific to sexual coercion, including impulsiveness, general hostility, and misogynous beliefs (i.e., hostile attitudes toward women). The impersonal sex path is also hierarchical and involves early family violence, child sexual abuse, adolescent delinquency, and sexual promiscuity.

In addition, persistent sexual coercers reported more alcohol consumption in sexual interactions than did the other groups. Persistent sexual coercers also exhibited more hostile attitudes about women and alcohol consumption in sexual interactions than did men who were not initially sexually coercive but initiated sexual coercion over the 1-year follow-up. A limitation of this study was the small number of initiators (n = 11) and persistent coercers (n = 17). Moreover, all the men were from a single university, and self-report was the sole means of assessment. Although one third of the sample was not European American, there was no investigation of ethnic group differences or possible cultural influences.

There exists a dearth of knowledge on the role of cultural influences in the course of sexual coercion. Although some general influences, such as parenting, peer, and neighborhood processes, have been identified in the development of aggressive behavior in ethnic minority communities, cultural influences have been understudied. Hall and Barongan (1997) hypothesized that there are culture-specific influences that may moderate the situational and developmental risk factors for sexual coercion proposed in the Hall and Hirschman (1991) model. For example, an emphasis on interpersonal harmony in collectivist cultural groups may be a deterrent against acts that would upset interpersonal harmony, such as sexual coercion. In support of a cultural protective model, Hall, Sue, and colleagues have found that loss of face, a culturally based construct that involves a concern about fulfilling one's social role and the impact of one's behavior on others, is a protective factor against sexual coercion among Asian American men but not among European American men. However, these two studies were cross-sectional, and the effects of culturally based constructs on the course of sexual coercion are unknown.

Another general limitation of the research on the course of sexual coercion and of sexual coercion research more generally is that the dependent measure in most studies has been self-report of the perpetrator or of a significant other. The advantage of self-report is that sexually coercive behavior is typically a private event that is not detected by authorities and may often be unknown to anyone other than the perpetrator and the victim. However, the perpetrator may be motivated to dissimulate for reasons of social desirability. A significant other may be able to corroborate or dispute the perpetrator's self-report, but the significant other may not be aware of all the perpetrator's sexually coercive behavior, particularly when there are multiple victims involved that are unknown to the significant other.

In an effort to supplement self-reports of sexual coercion, Hall and Hirschman (1994) developed a laboratory analogue of sexual harassment. Participants who admitted to engaging in multiple acts of sexual coercion and others who admitted to no sexual coercion viewed films that included forced sex and violence and a neutral film. It is probable that some of the men in the sexually coercive group were chronic sexual coercers. Participants then were asked to select and show one of these films to a female confederate. Although the confederate was not depicted as disliking either film, the act of showing a film that contained sex and violence was considered harassing because of the film's noxious content. Fifty-two percent of the sexually coercive men showed one of the films that included forced sex and violence, whereas 8% of the noncoercive men showed one of these films. Moreover, those who showed one of the films that included forced sex and violence reported that the confederate was more upset by the films than did those who showed the neutral film, although the confederate's responses to both films were neutral. Insofar as the film-showing procedure involved the presentation of sexually coercive stimuli deemed noxious to the female confederate, it is an analogue of sexually harassing behavior that can be observed in the laboratory.

College populations are informative in the study of the course of sexual coercion because college is a time of increased independence and allows increased opportunities for social, sexual, and sexually coercive contact. In a longitudinal study that followed a large cohort of 835 male freshman over 4 years of college, incidence rates of sexual coercion ranged from 17% to 34% per year, although it is unclear from this study how many men were not previously sexually coercive and initiated or how many desisted in any particular year. Although college samples are not necessarily representative, they may provide a better perspective on the course of sexual coercion than clinical or forensic samples, which involve only those who have been apprehended for sexually coercive behavior and have been the primary focus of sexual coercion research. Consistent with prior theory and research, we hypothesized that persistent sexual coercers would exhibit greater deviance than noncoercers or than desisters and initiators whose sexual coercion is more transitory.

We modified the Hall and Hirschman (1994) laboratory procedure in the current study by presenting male participants, before they viewed or showed the films, with information that the female confederate strongly disliked sexual material in the media. Having this information and showing the sexual film to the female confederate is analogous to situations in which lack of consent to sexual contact is clearly communicated but ignored by a sexual coercer. We hypothesized that persistent sexual coercers would be more likely than others to show a sexual film to the female confederate. Persistent sexual coercers are more likely than others to be sexually coercive across contexts, and the laboratory analogue provides a context in which to engage in a form of sexual harassment.

The current study involved Asian American and European American men because of the differences between these ethnic groups identified in previous research by Hall, Sue, and colleagues. We hypothesized that cultural factors would protect against sexual coercion among Asian Americans and that cultural factors would be less influential among European Americans. Specifically, we hypothesized that those Asian Americans who have a stronger cultural identity would be less likely to be sexually coercive than those having a weaker cultural identity.

In previous work, Hall et al. (2005) found that cultural factors attenuated the effects of risk factors for sexual coercion. These attenuation effects were stronger for Asian Americans than for European Americans. On the basis of these findings, we hypothesized that cultural factors would attenuate the effects of risk factors among Asian Americans.

Discussion

Group Differences

Consistent with our hypothesis, the current results generally suggest that persistently sexually coercive men are different than men whose sexual coercion is more transitory or men who are not sexually coercive. Persistent coercers were higher than the other men in risk factors for sexual coercion, including delinquency and hostile masculinity. Our results lend support to Malamuth et al.'s (1991) conceptualization of these variables as risk factors for sexual coercion. The only variable on which the persistent sexual coercers did not exceed the other groups was early family violence. It is possible that this failure of early family violence to distinguish among the groups is because the sexual coercion groups in the current study were defined by behavior over 1 year, whereas in previous studies in which early family violence was associated with sexual coercion, the time frames were longer than a year. The results of this study suggest that recent delinquency and current attitudes may be better predictors of sexual coercion within the past year than distal abuse experiences.

Those men who desisted from sexual coercion were the largest group of sexual coercers, which is consistent with past research, and comprised 21% of the current sample. However, the follow-up period in the current study was 1 year, and it is possible that such desistance is only temporary. Also, 7% of the sample initiated sexual coercion during the course of the study, and 9% indicated sexual coercion before the first assessment as well as during the year prior to the second assessment.

Similar to Abbey and McAuslan's (2004) findings, our findings indicate that men who initiated or desisted from sexual coercion fell between the noncoercers and persistent coercers on most measures. Unlike in Abbey and McAuslan's (2004) study, in which initiators and desisters were significantly different from noncoercers and persistent coercers on several measures, initiators and desisters were not different in the current study. The absence of differences in the current study may be a function of the larger sample size and variability within the sexual coercion pattern groups, which included 2 to 4.5 times as many participants per sexual coercion pattern group relative to the Abbey and McAuslan (2004) study. It is possible that greater variability within the initiator and desister groups in the current study attenuated group differences. It is also possible that these men are part of a sexually coercive subtype that is not deviant on personality measures. Additional research is necessary to determine whether initiators and desisters are less deviant than persistent coercers or more deviant than noncoercers.

The group comparisons revealed both ethnic differences and similarities. Although Asian Americans had higher scores than European Americans on the risk factors of early family violence, acceptance of violence, and hostile masculinity, there were no ethnic differences in distributions of sexual coercion patterns or in laboratory sexual harassment. European Americans exceeded Asian Americans only on the risk factor of delinquency. Thus, Asian Americans were not at greater risk for actual sexual coercion than European Americans, despite the presence of more risk factors. This finding is consistent with past research in which differences in the frequency of sexual coercion between Asian and European Americans were not found.

Of the culturally relevant variables, loss of face was predictive of ethnic group differences and accounted for variation in risk factors evidenced by the attenuation of ethnic differences on risk factors when loss of face was included in the analyses. Loss of face was a protective factor against delinquency, acceptance of violence, and hostile masculinity for Asian Americans. However, ethnic identity was not associated with the ethnic group differences, nor was it generally associated with sexual coercion. In previous work by Hall et al. (2005), ethnic identity also was not directly associated with sexual coercion, although it did interact with other variables associated with sexual coercion.

Prediction of Sexual Coercion and Harassment

The strongest predictor of self-reported sexual coercion over a 1-year follow-up was past self-reported sexual coercion. A history of sexual coercion also is a relatively strong predictor of sexual coercion in offender populations. The risk factors, particularly delinquency, were also significantly predictive of self-reported sexual coercion.

Perceived minority status was significantly predictive of self-reported sexual coercion both for Asian Americans and European Americans. Hall et al. (2005) previously found perceived minority status to be associated with European American men's self-reported sexual coercion in a cross-sectional study. A theoretical explanation for ethnic minority men is displacement of the negative effects of minority status, particularly discrimination, in the form of abusive behavior onto women. It is possible that European American men who perceive themselves as the target of discrimination also engage in displacement. Although perceptions of discrimination in a majority group versus a minority group may also be qualitatively different, there is evidence that European Americans are aware of their minority status in certain situations.

A finding specific to Asian Americans in this sample and consistent with cross-sectional data was that loss of face attenuated the risk for self-reported sexual coercion created by the risk factors. Concern about fulfilling one's social role and about losing face when failing to do so is more prominent in Asian American cultures than in European American ones. Loss of face may be a protective factor against risk for sexual coercion among Asian American men to the extent that prescribed social roles do not include sexual coercion.

The sexual film showing analogue of laboratory sexual harassment corresponded with self-reported sexual coercion, which supports the external validity of the laboratory procedure. Loss of face attenuated laboratory sexual harassment for both Asian Americans and European Americans. Loss of face for European Americans attenuated a form of sexual harassment in a public setting (i.e., the laboratory) in which behavior is being monitored by an experimenter, but it did not attenuate the self-report sexual coercion measures, which were completed confidentially. This finding is analogous to data that 35% of a sample of primarily European American college men who were not sexually coercive reported that they would engage in sexual coercion if they could be assured of not being caught. Being caught involves detection of sexual coercion in a public setting, whereas not being caught involves undetected or private sexual coercion. Thus, loss of face may have situation-specific effects among European Americans but is a more general deterrent against sexual coercion among Asian Americans, for whom loss of face was a protective factor against both self-reported sexual coercion and laboratory sexual harassment.

Strengths and Limitations

The study has several strengths and limitations. First, one of the important features of the investigation was the longitudinal aspect, which allowed us to identify noncoercers, desisters, initiators, and persistent coercers as well as predictors or correlates of these types. Second, we were able to examine the importance of cultural variables in sexual coercion. Because ethnic research has been criticized for using ethnic labels without establishing the psychological meaning of ethnicity, we examined psychocultural constructs (e.g., loss of face) along with ethnicity. Thus, it was possible to unpackage the meaning of ethnicity. Third, the study included both self-report measures and a laboratory-based behavioral indicator of harassment. Finally, the sample sizes were substantial enough to permit various within-group analyses.

A limitation of the current study is that most of the participants were in college. Moffitt (1993) hypothesized that most chronic offenders would not develop the social and academic skills required for entry into college, and such chronic offenders are likely to be underrepresented in this sample. Nevertheless, psychopathy characteristic of chronic offenders has been identified in college students, and psychopathy in college students is associated with sexually coercive behavior. Although the persistent sexual coercers in the current sample were a relatively small group, they were more deviant than the other groups in the study.

Another limitation is the high rate of attrition in the sample from Hawaii. Nevertheless, those who were assessed at the follow-up differed from those who dropped out of the study on few variables, and these differences were small in magnitude. Completers and participants lost to attrition were not significantly different in sexually coercive behavior.

A further limitation of this study is the relatively short follow-up. It is possible that the patterns of sexual coercion identified in these results would change with a longer follow-up period. There is evidence of variation in the incidence of sexual coercion beyond 1-year periods, which implies that persistence, desistence, and initiation may vary over periods longer than 1 year.

Conclusions and Clinical Implications

The current results reveal a subtype of men that engages in persistent sexual coercion and others for whom sexual coercion is more transitory. These persistent sexual coercers are characterized by adolescent delinquency and attitudes accepting of violence and misogyny. Interventions with these men are less likely to be successful than with men whose sexual coercion is more transitory.

The most common form of sexual coercion in this study was coercion that desisted. Although such sexually coercive behavior may desist without treatment interventions, treatment interventions may expedite desistence and perhaps decrease the likelihood of relapse. Moreover, it is possible that such desistance may be transitory without treatment interventions.
The current results also have prevention implications. The strongest predictor of sexual coercion in this study was past sexual coercion, and men who had been sexually coercive at the first assessment were nearly eight times as likely as those who had not been sexually coercive to show recidivist behavior during the 1-year interval until the second assessment. Thus, preventing sexual coercion from initially occurring would substantially decrease the risk of establishing a persistent pattern of sexual coercion. Interventions with delinquent adolescents and those having attitudes accepting of violence and misogyny also could reduce the risk of sexual coercion. Moreover, interventions may be appropriate with adolescents and preadolescents to prevent these attitudes and delinquent behaviors from developing.

The current results suggest that preventive interventions not only with adolescents and preadolescents but with all college-age men may be needed. Forty percent of the men in this sample were sexually coercive before or during the course of the study, including 7% who initiated sexual coercion during the study. Such interventions may be targeted at reducing or preventing the development of acceptance of violence and misogynous beliefs. The current results also suggest that culture-specific interventions may be useful. Loss of face was a protective factor against sexual coercion for Asian Americans but not for European Americans. For Asian American men, such interventions might involve enhancing awareness of culturally prescribed social roles and how to fulfill them, as well as enhancing awareness of cultural sanctions for failing to fulfill them.