The sexual coercion literature has suggested numerous factors related to aggressive sexual behavior. The present investigation explores a number of these factors in a community sample. Data collected from 189 volunteers from the community included measures of sexual arousal, social perception, personality variables, attitudes toward women, and self-reported likelihood to rape. Multiple-regression analyses were used to determine the relative association of these factors to coercive sexual behavior. The present findings suggested that social perception, Extraversion and Neuroticism from the Eysenck Personality Inventory, sexual arousal, and self-reported likelihood to rape all contributed to the multiple regression. Rape Myth Acceptance, although not contributing significantly to the multiple regression, did show a significant zero-order correlation with coercive sexual behavior.
Additional analyses were performed in an attempt to replicate an earlier predictive study by Malamuth and Check (1983) that found self-reported sexual arousal to be predicted by a combination of self-reported likelihood to rape, Psychoticism and Neuroticism from the Eysenck Personality Inventory, power motivation, and sexual experience. In the present study, both self-reported sexual arousal and penile tumescence measures were significantly related to attitudinal measures, social perception measures, and self-reported likelihood to rape. Limitations of the present study and suggestions for future research are discussed.
There have been several attempts in the literature to delineate factors related to male aggressive sexual behavior. The rationale for such research has risen from two somewhat separate, although not necessarily mutually exclusive, theoretical models. The first, which will be labeled the psychopathology or individual difference model, has clinically focused on individuals who have been convicted of rape. This model is reflected in various state sexual psychopath laws (Rada, 1978) that view rape not only as a crime but also as a symptom of a psychiatric disorder. This model has led to numerous studies comparing rapists to nonrapists on a variety of psychological tests (see Rada, 1978, for review), behavioral models focusing on deviant sexual arousal of rapists (Abel, Blanchard, &: Becker, 1978), and clinical classification systems based on psychodynamic formulations (Groth, 1979).
Much of this literature, however, has been criticized by feminist writers (Albin, 1977; Brownmiller, 1975), who view rape as an aggressive rather than a sexual crime and a natural extension of the male socialization process. Such socialization leads males to feel dominant to submissive females, with females perceived as sexual objects for men's pleasure. In this model, male sexual aggression is not the result of any individual psychopathology but is due to attitudes toward females and aggression learned as part of enculturation. Such a model would also include variables such as male dominance, perpetuation of traditional sex roles, and power differential between males and females. From this perspective, rape as legally defined represents only one small part of the sexual aggression experienced by women.
This perspective has led to numerous studies, mainly with college students, on the association between various attitudes (Sex Role Stereotyping, Rape Myth Acceptance, and Acceptance of Violence) to various measures of male sexual aggression (see Malamuth 8c Donnerstein, 1984, for an excellent review of much of this work). Incidence studies indicate that sexual coercion and rape occur much more frequently than official statistics indicate. Malamuth (1984) found, across a number of studies by his research group, that 35% of the males reported at least some (2 or greater on a 5-point scale) likelihood to rape if they thought they would not get caught. Although this self-report does not mean that such individuals would rape, other data suggest that a number of males do admit to at least some level of sexual aggression against women. Kanin (1969) found that 25.5% of 241 college males admitted to sexual aggression against women, and Koss, Gidycz, and Wisniewski (in press) found that 25% of college males admitted to sexual aggression. Similarly, although not directly indicating number of offenders, surveys also suggest frequent unreported sexual aggression against women. Russell (1982), in a survey of 930 women, found that 44% reported being victims of rape or attempted rape, although only 8% had reported such incidents to the police. Other surveys of females (Kanin & Parcell, 1977; Kilpatrick & Kanin, 1957; Koss & Oros, 1982), which have conceptualized sexual coercion as being on a continuum ranging in severity from such behaviors as kissing a woman against her will to using force to have intercourse, have found that at least 50% of women are victims of some form of sexual coercion.
The two models above suggest a number of factors that should be associated with sexual coercion. The sociocultural model would clearly suggest that the presence of various attitudes toward females, such as acceptance of rape myths, sex role stereotyping, and males' need to be dominant and tendency to sexualize females, should be related to coercive sexual behavior to a greater extent than personality or sexual arousal variables. The psychopathology, or individual difference model, on the other hand, would posit factors such as sexual arousal to rape depictions or personality factors such as generalized anger, hostility, and social inadequacy (Groth, 1979) as being significantly related to coercive behavior. In the past, there has been a tendency to treat these two models as if they were mutually exclusive. However, more recent empirical literature (Malamuth, 1986; Malamuth & Check, 1983) has recognized that incorporating variables from both models may be necessary to understand sexual aggression directed against females. Prior to outlining the present investigation, the authors will review studies related to these factors.
SEXUAL AROUSAL TO RAPE DEPICTIONS
Psychophysiological assessments of sexual arousal via direct penile tumescence measurement have consistently indicated that rapists show greater sexual arousal to rape stimuli than do nonrapists, although the two groups show equivalent responses to depictions of mutually consenting intercourse (Abel, Barlow, Blanchard, & Guild, 1977; Abel, Blanchard, Becker, & Djenderedjian, 1978; Barbaree, Marshall, & Lanthier, 1979; Quinsey, Chaplin, & Upfold, 1984; Quinsey, Chaplin, & Varney, 1981). Abel and his colleagues (Abel, Becker, 8c Skinner, 1980) have also suggested that rapists have difficulties in inhibiting their aggressive sexual arousal as compared to nonaggressive arousal when instructed to and show an increase in the rape index (a ratio of the arousal to rape stimuli compared to non-rape stimuli) when attempting to exert such control. However, no statistical analysis was provided; therefore, it is unclear whether the changes observed in this study were significant or whether the results reflect changes in aggressive arousal or arousal to nonaggressive cues.
Although the above studies are consistent in their findings, more recent data suggest that men who have not been convicted of rape also respond sexually to rape depictions. For example, Barbaree and his colleagues found that males fail to inhibit sexual arousal to rape stimuli when intoxicated (Barbaree, Marshall, Yates, & Lightfoot, 1983) or when angered by a female confederate (Yates, Barbaree, & Marshall, 1984) compared to conditions when they are not intoxicated or angered. In addition, Malamuth and his colleagues found increased sexual response to rape depictions when the victim is portrayed as being "sexually aroused" as measured by self-report (Malamuth & Check, 1980b; Malamuth, Heim, & Feshbach, 1980) or as assessed by penile tumescence measures (Malamuth & Check, 1980a). Also, Check and Malamuth (1983) found evidence that males high in sex role stereotyping respond to rape themes in a manner similar to that of convicted rapists and that selfreported likelihood to rape correlates with sexual arousal to rape depictions (Malamuth & Check, 1980a; Malamuth, Haber, & Feshbach, 1980). Finally, Malamuth (1983) found that sexual arousal to rape themes was also predictive of laboratory aggression against females and that arousal to rape stimuli was one factor related to self-reported aggressive sexual behavior (Malamuth, 1986).
These results suggest that although identified rapists do consistently show increased arousal to rape stimuli, a rather significant proportion of normals also show elevated sexual response to such material. In college student populations, at least, this arousal to rape material also appears related to sexual aggressive behavior. The generalizability of these results from identified rapists and college students to a sample from the general population is unknown. However, given the consistency of findings, similar results should be obtained in a nonclinical, non-college-student sample. In addition, the arouse versus suppress instructions that are generally used in clinical studies of rapists (Abel, Becker, & Skinner, 1980; Abel, Blanchard, Becker, & Djenderedjian, 1978) have not been used in general community samples. Abel et al. (1980) indicated that rapists show increased response to rape themes when attempt ing to control such arousal, although, as noted, these data must be interpreted cautiously. If true, the collection of sexual arousal measures under suppress instructions (that is, instructions to control their arousal) with "normals" may be related to sexual coercion.
Numerous early studies (e.g., Rada, 1978) attempted to describe the personality characteristics of rapists. Many of these early reports used projective tests with contradictory results. This is not surprising, given the problems in reliability and validity associated with projective techniques. Recent studies using objective personality inventories such as the Minnesota Multiphasic Personality Inventory (MMPI) (Armentrout & Hauer, 1978; Panton, 1978; Quinsey, Arnold, & Pruesse, 1980) have shown more consistent results: As a group, rapists show elevations on the Psychopathic Deviant Scale (Scale 4) and the Schizophrenic Scale (Scale 8) of the MMPI. The clinical interpretation of such elevations would indicate an individual who is antisocial in nature with significant feelings of insecurity, anger, and hostility. However, it should be noted that this profile is also observed in many other forensic psychiatric populations (Quinsey et al., 1980).
A number of studies have also looked at personality traits in nonrapist college samples and the relationship of such variables to aggressive sexual behavior. Rapaport and Burkhart (1984) found that the Responsibility and Socialization scales of the California Personality Inventory were negatively correlated with self-reported coercive sexual behavior among college males. Rapaport and Burkhart point out that "low scores on this scale reflect personality characteristics of immaturity, irresponsibility, and the lack of social consciousness and have traditionally been associated with delinquency and antisocial behavior." Using the Eysenck Personality Inventory, which has been used previously in studying sexual behavior in general (Eysenck, 1976), Malamuth and his colleagues (Barnes, Malamuth, & Check, 1984b; Malamuth & Check, 1983) have shown significant correlations between the Psychoticism and Neuroticism subscales of the Eysenck Personality Inventory and sexual arousal to rape depictions in a college student sample. High scores on Neuroticism and Psychoticism would suggest an aggressive, hostile, emotionally labile individual who is lacking in empathy. Similarly, Barnes, Malamuth, and Check (1984a) have found that Psychoticism correlates with more unconventional sexual behaviors, acquisition of sexual knowledge later in life, the engagement of sex more for power than for expression of love and affection, and a positive attitude toward the use of force in sexual activities. Extraverts, on the other hand, tended to engage in sexual activity for hedonistic reasons and for novelty, and Extraversion was positively correlated with conventional sexual behaviors. Neuroticism showed less correlation with sexual behaviors in general. These data suggest that the scales of the Eysenck Personality Inventory are related to factors that may be predictive of coercive sexual behavior.
Although feminist theory would not postulate personality or psychopathology as necessarily being associated with sexual aggression, both the clinical literature and the literature with college students suggest that there are individual differences (Malamuth & Check, 1983) among personality factors that are associated with aggressive sexual behavior. Across studies, factors such as antisocial traits, hostility, anger, and lack of empathy consistently emerge as factors correlated with aggressive sexual behavior. For the present study, it is hypothesized that similar personality traits or individual differences will be found related to coercive sexual behavior.
RAPE-SUPPORTIVE ATTITUDES ABOUT FEMALES
As reviewed (Albin, 1977; Brownmiller, 1975), feminist writers have proposed that attitudes held by males toward females and toward aggression are major causative factors in male sexual aggressive behavior. To date, the majority of studies have been limited to specific attitudinal variables such as Rape Myth Acceptance, Sex Role Stereotyping, and various aggressive attitudes, with less focus on other variables such as males' need to be dominant in relationships [see Malamuth (1986) and Malamuth & Check (1984) for exceptions]. In general, there has been significant support for this contention in studies among college students. Malamuth and Check (1980a), Malamuth, Heim, and Feshbach (1980), Malamuth, Haber, and Feshbach (1980) and Tieger (1981) have all found that subjects who self-report some likelihood to rape accept more rape myths. Malamuth (1983), in a study employing path analytic techniques, found that in addition to sexual arousal to rape depictions, attitudes facilitating violence (Rape Myth Acceptance and Acceptance of Interpersonal Violence scales developed by Burt, 1980) were predictive of laboratory aggression against women. As previously reviewed, Check and Malamuth (1983) also found that subjects high in sex role stereotyping showed sexual arousal patterns similar to rapists.
A number of studies published following the inception of the current study have also investigated factors related to actual self-reported coercive behavior in samples of college students. Rapaport and Burkhart (1984), in a study of personality and attitudinal factors related to coercive sexual behavior, found that Adversarial Sexual Beliefs and Acceptance of Interpersonal Violence (from Burt, 1980) and the Endorsement of Force scale (a scale designed specifically for that study to assess subjects' beliefs about the use of force in social/sexual situations) correlated with self-reports of coercive sexual behavior. However, the authors did not find more general attitudes such as sex role stereotyping, sexual conservatism, sex role satisfaction (Burt, 1980), or the Attitudes toward Women Scale (Spence, Helmreich, & Stapp, 1973) to be related to aggressive behavior. Koss, Leonard, Beezley, and Oros (1985) found that a number of rape-supported beliefs, many based on scales developed by Burt (1980), discriminated groups of men varying in degree of sexual coercion based on the Koss and Oros (1982) Sexual Experience Survey. Similarly, Malamuth (1986) found that, in general, sexual arousal to rape stimuli, dominance as a sexual motive, hostility toward women, acceptance of interpersonal violence, and sexual experience were significant predictors of coercive sexual behavior as measured by the Sexual Experience Survey. Malamuth also found that interactions between these variables improved prediction over a straight additive model. Ageton (1983), in a study based on a national probability sample of adolescents, found attitudes toward sex roles and attitudes toward rape and sexual assault as predictive of self-reported sexual aggression. However, when the variable Involvement with Delinquent Peers was entered, the measures of sex roles and attitudes toward rape explained few additional variables. This might be interpreted to suggest that rape-supportive beliefs (at least those measured in the Ageton study) are part of a more general disregard for the right of others, although further studies are needed in this area.
For college student samples, rape-specific and aggressive attitudes consistently relate to various measures of aggressive sexual behavior, whereas more general attitudinal variables, such as sex role stereotyping, show some inconsistency across studies. However, studies by Check and Malamuth (1983) and Rapaport and Burkhart (1984) also differ in the outcome variable, i.e., prediction of sexual arousal versus prediction of actual self-reported sexual behavior, which may explain the results. Similarly, although Koss et al. (1985) and Rapaport and Burkhart (1984) both used measures of self-reported coercive behaviors, they used different measures of sex role stereotyping.
A final variable, which may be related to actual coercive sexual behavior but which has received little attention in the literature, is males' perception of female behaviors. The authors' clinical experience with rapists suggests that many label female behavior as either hostile or seductive. Although such labeling may serve as a justification for their behavior, it may also represent deficits in their ability to separate seductive from friendly behavior or hostile from assertive behavior.
The sociocultural model hypothesizes that males tend to sexualize females' behavior and would support the notion of differences in perception of female behavior between coercive and noncoercive males. Evidence exists that males may misperceive or mislabel female friendly behavior as seductive, at least as compared to females' labeling of such behavior (Abbey, 1982; Giarrusso, Johnson, Goodchilds, & Zellman, 1979). In a college sample, Abbey found that both male observers and male actors in a heterosocial interaction tended to label the female actor as more promiscuous and more seductive than did either the female observer or the female actor. Similarly, Giarrusso et al., in a study of adolescents, found that males tended more often than females to perceive both situational cues (i.e., being in a guy's house alone) or behavioral cues (talking a lot about sex, playing with the other person's hair, etc.) as indications that the other person wanted to have sex.
In general, although limited data are available, it appears that males and females differ in their perception of sexual cues. Whether such misperceptions or mislabelings actually lead to sexual coercion is unknown. To the extent that rape is an aggressive and hostile crime (Groth, 1979) and that individuals who engage in aggressive sexual behavior are characterized as being hostile and angry, the misperception of hostile cues in females may also lead to actual coercive behavior. Measures of perceptions of hostility have not previously been investigated.
The previous review suggests a number of problems with the existing literature. Previous findings suggest that measures of attitudes toward women, sexual arousal to aggressive stimuli, personality factors, and social perceptions are correlated with selfreports of sexually aggressive behavior. This provides at least partial support for variables derived from both a sociocultural model and psychopathology or individual differences model. However, to date, the majority of studies have been with either college student samples, clinical samples of rapists, or adolescents. The generalizability of the above findings to a more general community sample is unknown. In addition, until recently (Ageton, 1983; Malamuth, 1986; Koss etal., 1985; Rapaport & Burkhart, 1984), the majority of studies have not attempted to relate such factors to actual self-reported coercive behavior. It may well be that the factors that predict either self-reported likelihood to rape if one would not get caught or those that predict sexual arousal to rape depictions are not predictive of actual behavior.
The present study was designed to address some of these issues by investigating relevant attitudinal measures, personality variables, and sexual arousal variables in a community sample. A new measure, one of social perception, was included to determine its relationship to coercive sexual behavior. For the purpose of this study, coercive sexual behavior was defined similarly to Kilpatrick and Kanin (1957) and represents subjects' self-reported frequency of four increasingly coercive sexual behaviors, from kissing a woman against her will, to touching a woman's breasts against her will, to touching a woman's genitals against her will, to forcing a female (or male) to do something sexual. It was hypothesized, based on previous literature and clinical experience, that variables drawn from the sociocultural model (Rape Myth Acceptance, social perception, etc.) and psychopathology model (sexual arousal to rape depictions and personality) would correlate with coercive sexual behavior. In an attempt to clarify inconsistencies in previous studies, the sexual arousal measurement procedures were expanded by collecting erection responses under different instructional sets (e.g., arouse and suppress instructions).
Finally, a series of analyses are presented, which partially replicate Malamuth and Check (1983). This permits some comparisons between the community sample in the present study and the college student samples in earlier studies. Also, it allows some determination of whether factors that relate to self-reported coercive behavior are similar to those that relate to sexual arousal to rape depictions.
The goal in subject recruitment was to obtain a sample of males that would be representative of the Memphis, Tennessee, community. An attempt was made to recruit across socioeconomic levels, ethnic backgrounds, and age. To accomplish this, flyers describing the program were sent to personnel managers at local industries for posting, to prison guards, and to union representatives, and were posted in hospitals surrounding the research project. Subjects were also recruited through personal contacts of the staff and from friends of study participants, who were asked to tell associates about the study. The flyers instructed subjects to phone for an appointment, and 213 subjects were seen for an initial screening interview. Two subjects were excluded after this initial interview as a result of questionable mental status. In addition, because the stimuli for the physiological assessments were heterosexual in nature, three subjects who were predominantly homosexual were also excluded from the sample. The mean age of the remaining 208 subjects was 31.5 years (range 18 to 69). Demographic information compared with Memphis population statistics is presented in Table 1.
All subjects were paid $5 per hour for participation, with a maximum fee of $35 for completion of all components of the study. Subjects were informed that if they chose to withdraw from the study, they would be paid the money earned up to the point of withdrawing. All subjects agreed to participate, and no subject withdrew after beginning the study.
Comparisons to the Memphis population suggest that the present sample has a slight overrepresentation of blacks as compared to Caucasian subjects and is younger than the general population. Also, subjects in the present sample have more years of education, although economic level is lower. The lower economic level probably results from at least two factors: (a) the sample is slightly younger; therefore, present earnings have not reached their potential; and (b) some of the subjects were skilled blue-collar workers who had been employed in large manufacturing plants that had closed or drastically reduced their labor force during the time of the study. Although the final sample does not represent the Memphis population, there is a greater diversity in age, educational background, income levels, and ethnic background than in the college student samples previously reviewed.
Materials and Apparatus
Sexual arousal measures. Zuckerman (1971) has indicated that direct measurement of penile tumescence appears to be the most reliable and valid physiological measure of male sexual arousal. For the present study, direct measurement of penile circumference was performed by using a metal band strain gauge. All responses were recorded via a Grass Model 7 polygraph equipped with appropriate preamplifiers. Rape and heterosexual videotaped stimuli used were those developed by Abel, Blanchard, Becker, and Djenderedjian (1978), and lesbian videotapes were employed to control for novelty.' The rape stimuli depicted simulated rape that clearly showed the male using force, and the heterosexual films depicted simulated intercourse between a male and a female with the female initiating the sexual behavior. The lesbian film depicted two females engaged in a variety of sexual behaviors and was a segment from commercially available pornography. Heterosexual and rape films depicted black confederates for black subjects and Caucasian confederates for all other subjects. A decision was made to use Caucasian films for the small number of subjects who were in the other racial groups (which included American Indian and individuals of Oriental descent) because no stimuli were available that used the same race models for those two groups.
Personality variables. The Neuroticism, Extraversion, and Psychoticism scales of the Eysenck Personality Inventory were used as measures of personality function. These scales evidence good test-retest reliability (.7 to.9) across studies (Eysenck & Eysenck, 1975). As noted, the scales have previously been used in studying sexual behavior (Eysenck, 1976) and specifically in studying unusual and aggressive sexual behavior (Barnes, Malamuth, & Check, 1984a, 1984b).
Rape-supportive attitudes about females. The scales developed by Burt (1980) measuring Rape Myth Acceptance, Sex Role Stereotyping, Adversarial Sexual Beliefs, Sexual Conservatism, and Acceptance of Interpersonal Violence were employed to measure attitudes supportive of coercive sexual behavior. Items for these scales are rated from 1 (Agree) to 7 (Disagree), except for six questions on the Rape Myth Acceptance scale. These six items asked subjects to rate the extent to which they would believe the reports of various individuals (a White woman, a Black woman, a young boy, etc.) as to being victims of rape on a 1 (Always) to 5 (Never) scale. Psychometric data presented by Burt indicate that these scales are internally consistent, with Cronbach alphas for all of the scales being .80 or greater. As noted in the introduction, these scales have been shown to be related to various measures of sexual aggression across a number of studies.
Social perception. Social perception in the present study was defined as the subject's ability to discriminate friendly from seductive behavior and assertive from hostile behavior in females within a male/female interaction.
Subjects viewed 20 30-s videotapes depicting four different male/female interactions. In each scene, a male actor attempted to initiate a social interaction with a female. The female actor responded in one of four ways: hostilely, assertively, seductively, or pleasantly. In an attempt to ensure reliability, each scene was repeated five times with different actors, so that each subject viewed a total of 20 stimulus scenes presented in a random order.
Two sets of social perception scenes were developed so that black subjects viewed black actors, and non-black subjects viewed Caucasian actors. Actors ranged in age from their mid-20s to mid-3Os and were graduate students in psychology and/ or colleagues of the investigators. Ten blacks (five males and five females) and ten whites (five males and five females) participated as actors.
Subjects rated the scenes using a 7-point semantic differential on 14 adjectives. Three critical adjectives (seductive, rejecting, and hostile) were chosen a priori for analysis. These adjectives were chosen on the basis of investigators' clinical experience with rapists who tend to label women as seductive, hostile, or rejecting. The remaining 11 adjectives (assertive, appropriate, sexy, unfriendly, pleasant, fair, socially skilled, likable, sympathetic, tactful, and hard to get) were included so that subjects could not easily interpret the hypothesis of the study. The data, then, consisted of three ratings from each of 20 scenes, with higher values indicative of the critical adjectives-seductive, rejecting, and hostile.
A previous pilot study (Murphy, Coleman, Haynes, & Flanagan, 1982) with seven subjects who were mental health professionals or individuals rated as highly socially skilled, indicated the subjects could reliably discriminate between these different scenes. In this pilot study, the mean for the hostile adjective for the hostile scene was 6.31 (S.D. = .87); for the assertive scene, the mean for the adjective was 3.46 (S.D. = .95) (p <.001). Similarly, for the seduction adjective, the mean for the seduction scene was 6.06 (S.D. = .67), compared to 4.49 (S.D. = .67) for the friendly scene (p <.01).
Self-reported likelihood to rape. As stated earlier, the index for self-reported likelihood to rape has been shown to relate to several factors examined in the present study. In this study, subjects were asked to rate on a 9-point scale (1 = 0, 9 = 100) the likelihood that they would rape if they could be assured that no one would know and they could in no way be punished. The 9-point scale was used rather than the 5-point scale usually employed by Malamuth and his colleagues because it was felt at the time that it might give a larger distribution of scores.
Coercive behaviors. The four coercive behaviors measured in this study, as noted, were similar to those described by Kilpatrick and Kanin (1957). The questions asked were how many times have you (a) kissed a woman when she didn't want you to; (b) touched a woman's breasts when she didn't want you to; (c) touched a woman's genitals when she didn't want you to; (d) forced a female (or male) to do something sexual that she (he) didn't really want to do.
General procedures. Initially, all subjects were given a written consent to read while it was read to them by an investigator. The consent form described the study as a research project of male/female social and sexual relationships. The physiological measurement procedures were described in detail and included showing the patient the penile transducer. Confidentiality was guaranteed, and subjects were told that they were free to withdraw from the study at any time. After signing the informed consent, subjects participated in a brief semistructured interview focusing on sexual history, drug and alcohol history, and general psychological functioning. Subjects were interviewed individually in a private office by one of the three investigators (one male, two females) each with more than 5 years experience in taking sexual/ psychosocial histories and in working with sex offenders. The remainder of the assessment was divided into four basic components: completion of questionnaires, completion of the social perception measure, and completion of the first and second sexual arousal measurements. No set order was used for completion of the components, except that the sexual arousal measurement session occurred at least 24 h following the first sexual arousal measurement session. Order was determined by technician and laboratory availability, and order of assessment was not recorded; therefore, analysis of order effect was not possible.
Following completion of all assessment components, subjects were debriefed by one of the investigators in a manner similar to that described by Malamuth and Check (1984), although no formal written debriefing script was followed. The debriefing stressed the horror of rape and the significant negative impact rape has on the victim.2 Donnerstein and Berkowitz (1981), Check and Malamuth (1984), and Malamuth and Check (1984) have found that such debriefing can be educational, leading to less acceptance of rape myths as compared to a control group not debriefed in such a fashion.
Sexual arousal assessment procedure. Initially, subjects sat in comfortable reclining chairs, the technician attached electrodes to measure skin resistance, respiration, and blood volume and instructed the patient on the placement of the penile transducer. The technician left the room, and subjects positioned the penile transducer in the privacy of the laboratory. All stimulus presentations were controlled by a technician in a separate room who was in verbal but not visual contact with the subject. Two technicians, a male and a female, each having more than 10 years experience in the physiological assessment of sexual arousal, were employed in conducting laboratory sessions. Following each stimulus presentation, subjects rated their percentage of full erection, percentage of subjective arousal, and the degree to which they attempted to control their arousal. Interstimulus interval was a minimum of 2 min or until penile response returned to baseline, defined as the polygraph pen being within ±3 mm of the original baseline established prior to commencement of this session, when the subject reported 0% arousal. Only the percentage of full erection and, for the subanalysis, the percentage of subjective arousal were included. All erection responses were converted to percentage of full erection (Laws & Osborn, 1983). If 100% erection was not established during the exposure to the eight 2-min stimuli, subjects were then exposed to a longer (10-min) explicit heterosexual film following completion of all components of the study. If 100% erection was not achieved by this procedure, subjects were then asked to remove the transducer, stimulate themselves to a full erection and replace the transducer. Approximately 5% of subjects did not achieve a full erection using any of these procedures, and their full erection was estimated from their largest response and their self-reported estimate of that response. That is, if a subject showed a 15 mm pen deflection on the polygraph paper and reported that as a 50% erection, their full erection was estimated to be equivalent to a 30-mm pen deflection.
Eight 2-min stimuli (four rape, two heterosexual, and two lesbian) were presented in a pre-set random order except that a lesbian film was always presented as the first stimulus. Half of the stimuli in each category were presented under instructions not to interfere with arousal (labeled arouse instructions); the other half were presented under suppress instructions, as described by Abel and his colleagues, (Abel etal., 1980; Abel, Blanchard, Becker &: Djenderedjian, 1978). Four rape stimuli were used to avoid skewing the data by one artificially elevated response. These eight stimuli were presented on two occasions separated by at least 24 h. In the first session, a 3-min neutral (nature) film was presented immediately before sexual arousal measurement; in the second, 1 3-min film of a female insulting a male was employed. This latter tape was developed to elicit anger as part of a separate study.
Rape indices were developed by subtracting percentage of erection to rape stimuli from percentage of full erection to mutual stimuli (Malamuth, 1983). Indices were calculated separately for arouse and suppress trials and separately for the first and second laboratory sessions. Thus, there were four indices: arouse/Session 1 index, suppress/Session 1 index, arouse/Session 2 index, suppress/Session 2 index.
Self-report and questionnaire assessment procedure. The Burt scales, the self-reported likelihood to rape questions, and the questions related to frequency of coercive behavior were all embedded within a demographic sexual history questionnaire. 3 Subjects completed this questionnaire individually along with the Eysenck Personality Inventory, with a technician available to answer questions if necessary.
Social perception assessment procedures. For the social perception assessment, the subject sat in a comfortable reclining chair with a television monitor positioned approximately 4 ft directly in front of the subject. The technician initially instructed the subject on the use of the rating scale and then left the room. The technician, who was in verbal but not visual contact with the subject, controlled the presentation of the social perception tapes from an adjoining equipment room. The tapes were presented in random order and were stopped after each scene to allow the subject to complete the rating scale for that scene. Once the rating was completed, the next tape was presented.
Although the original goal of this study was to ascertain the relationship of a number of variables to self-reported coercive sexual behavior in a community sample, a number of preliminary analyses were first conducted. The purpose of these analyses was to assist in reducing the number of variables measured and to provide basic data regarding self-reported coercive behavior. To assist in variable selection, all of the variables for the present study were submitted to a principal component factor analysis with varimax rotation (BMDP) (Dixon, 1981). One hundred eighty-nine subjects who had complete data on all variables were included in this analysis. For brevity and clarity, only those results from the factor analysis relevant to variable selection will be reported. Full results of the factor analyses are available from the first author.
Sexual arousal. Since there were two laboratory assessment sessions, the physiological data were initially submitted to a two-factor (session x arouse/suppress) randomized block analysis of variance (ANOVA), with the rape indices as the dependent variable. The results indicated a nonsignificant session effect, which suggests no difference in responding between sessions.
Because there was potentially a large number of variables that could be derived from the physiological assessment, including the four rape indices previously described, plus the data on percentage of erection to the various stimuli, the factor analyses were used to reduce the number of variables. Results indicated that the rape indices loaded on one factor, and the percentage of erection to rape stimuli loaded on a second factor. For the indices, the arouse and suppress indices in Session 2 had the highest factor loadings. For actual percentage of erection to rape stimuli, the arouse and suppress responses during the first session had the highest loadings. As a result of the factor analysis, the arouse and suppress indices in Session 2 and the actual percentage of erection to rape stimuli under arouse and suppress instructions during Session 1 were used as four measures of sexual arousal in this study. The rationale for choosing these was the desire to include indices under both arouse and suppress instructions; it was felt that because erection measures loaded on a different factor (which has not been reported previously), both types of measurements should be included.
Rape-supportive attitudes about females. The factor analysis of the five attitudinal variables indicated that they all loaded heavily on one factor. Factor loadings were .792 for Sexual Conservatism, .782 for Adversarial Sexual Beliefs, .752 for Rape Myth Acceptance, .716 for Sex Role Stereotyping, and .629 for Acceptance of Interpersonal Violence. Because of the high interrelationship between these scales, only three were chosen for further analysis. Rape Myth Acceptance and Sex Role Stereotyping were chosen because of the empirical relationship they have shown to reactions to rape stimuli in college student populations (Check & Malamuth, 1983; Malamuth, 1981, 1983).
The Adversarial Sexual Beliefs scale-the expectation that sexual relations will be exploitive or manipulative and that the other party cannot be trusted-also was included because it seemed to reflect the verbal reports heard in clinical interviews with rapists during the principal investigator's 8 years experience in treating and evaluating rapists in both outpatient and inpatient settings. The recent findings of Rapaport and Burkhart (1984), previously reviewed, provide further justification for the use of this scale. However, a recent study (Malamuth, 1986), which was not yet published at the inception of this investigation, suggests that Acceptance of Interpersonal Violence might be a somewhat better predictor, although there appears to be a good deal of overlap between the variables.
Social perception. For data analysis, three discrimination indices were calculated from subjects' ratings. Subjects' ratings on the semantic differential did not differ across actors. Therefore, for each of the four types of stimulus scenes (hostile, assertive, seductive, pleasant), the subject's mean rating on each of the three critical adjectives was computed across the five repetitions. The hostility discrimination index was derived by subtracting the subject's mean rating on the hostility adjective for the hostility scenes from the subject's mean rating on the hostility adjective for the assertive scenes. Similarly, the seduction discrimination index was derived by subtracting the subject's mean rating on the seduction adjective for the seduction scenes from the subject's mean rating on the seduction adjective for the friendly scenes. A rejection discrimination index was derived by subtracting the subject's mean rating on the rejection adjective for the hostile scenes from the subject's mean rating on the rejection adjective for the assertive scenes.
Factor analysis indicated that the rejection and hostility discrimination indices loaded highly on the same factor (loadings = .788 and .779, respectively), and therefore only the hostility discrimination index and the seduction discrimination index were employed to help reduce the number of variables. In these indices, the smaller the value, the poorer the subject's ability to discriminate females' behavior.
Coercive behaviors. Because of the importance of self-reported coercive behaviors to this study, rather extensive basic data regarding this variable will be presented. Prior to this analysis, subjects were excluded who either admitted to previous arrests for sexual crimes (N - 5) and/or who did not report frequencies for all four coercive behaviors (N = 7). Subjects who reported previous arrests and convictions for rape were excluded (a) because of the small number of such individuals and (b) because they denied their guilt of the crime. Previous research (Murphy, Krisak, Stalgaitis, & Anderson, 1984) indicates that erection measurements in a population of incarcerated sex offenders denying the charges were unreliable. It was felt that such subjects might invalidate their data, and therefore a decision was made to exclude these individuals.
The means for the behaviors are presented in Table 2. Fifty-three percent of subjects reported kissing a woman against her will; 55% reported touching a woman's breasts against her will; 42% reported touching a woman's genitals against her will; and 17% reported forcing a female to do something sexual she did not want to do. AN OVA indicated a significant difference in the frequency of the four behaviors (F - 5.97, df= 3,585, p < .001), although the Newman Keuls post-hoc test indicated that only the force question was significantly different (p < .05) from the other three. In general, the coercive behaviors did decrease in frequency as proposed except that the frequency for kissing a woman against her will and touching of breasts against a woman's will were reversed.
Correlations between the four variables are presented in Table 2. Sample size for the analysis was 189, which represents all subjects who had complete data for all variables used in the regression analyses to be presented. The mean frequencies for each of the variables with this sample size are indicated in Table 2. The patterns of correlations suggest significant relationships between most of the behaviors, although the relationship between kissing and the other variables is quite small.
Finally, since it was proposed that the four coercive behaviors increased in level of coercion, this was tested through subjecting the scale to a Guttman scaling analysis (Nie, Hull, Jenkins, Steinbrenner, & Bent, 1975). An important aspect of the Guttman analysis for the present project was determination of whether the individual questions from the coercive behavior self-report formed a cumulative scale. That is, an individual who scores positive on the most coercive item (i.e., use of force) should score positive on the other coercive behaviors. For the purpose of the Guttman analysis only, each behavior was scored as a 1 if frequency was 1 or greater and as 0 if reported frequency was 0. Also, two Guttman analyses were formed: one using all four questions, the other using only three items (touching breasts, touching genitals, and using force). This was done because the item relating to kissing did not correlate well with the other variables, and the frequency of this variable was not as predicted. Analysis of the four-item scale produced a coefficient of reproducibility of .896, which represents the degree to which an individual's score on the scale is predictive of response pattern. For example, if the scale had a perfect coefficient of reproducibility, it could be predicted with 100% accuracy that a subject who scores 1 on the force question will also score 1 on the other questions in the scale. In general, a coefficient of reproducibility greater than .9 is required to consider the scale a valid Guttman scale (Nie et al., 1975). Similar analysis for the three-item scale produced a coefficient of reproducibility of .941, which is within the acceptable range. Because of these results and the fact that the kissing variable did not correlate well with the other behaviors, it was dropped from further analyses.
Factors Related to Self-Reported Coercive Behaviors
To determine the association of the previously described varibles to coercive behavior, the sum of the frequencies across the three coercive behaviors was calculated to produce a total coercion index, and the sum was entered as the dependent variable into a multiple linear regression (BMDP1R) (Dixon, 1981) along with previously described independent variables. Results are presented in Table 3 with variables listed by size of the standardized regression coefficient. The overall multiple R was .41 (R2 = .17) which was significant (F = 2.73, df= 13,p = .002). Also indicated in Table 3 is the simple Pearson product moment correlations of each of the predictor variables with the criterion (total coercion index). Comparison of the standardized regression weights to the simple correlations of the predictor with the criterion allows some assessment of the correlation of predictor and criterion variables independent of its correlation with other predictor variables. Table 4 presents the correlation matrix for all of the predictor variables to assist in interpreting the current multiple regression analysis.
Inspection of the standardized regression coefficients indicates that the hostility discrimination index and the Extraversion and Neuroticism scales were all significant predictors of self-reported coercive behavior and that the rape index under arouse instructions and the rape likelihood index approached significance. Inspection of the zero-order correlations again shows the hostility discrimination index and the rape likelihood index significantly correlated with the total coercion index, and the correlation between the rape index under arouse instructions approached significance. When inspecting the simple correlations, it is also observed that the Rape Myth Acceptance scale was significantly correlated with the criterion, and the Psychoticism scale of the Eysenck Personality Inventory approached significance. These two scales did not add to the multiple regression, possibly because of their shared variance with other variables already in the predictor set (see Table 4). On the other hand, Neuroticism and Extraversion showed no zero-order correlations with the criterion but were significant predictors in the regression equation. This suggests that they may have been serving as suppressor variables (Cooley & Lohnes, 1971). That is, the variable may have added to the regression equation because of a useful relationship to other predictors rather than because of a direct relationship with the criterion. However, inspection of Table 4 shows that the only significant correlations are between Neuroticism and Psycoticism and between Neuroticism and Extraversion. Therefore, it is difficult to determine clearly the impact of these variables.
In summary, among the independent variables considered, self-reported coercive sexual behavior could be estimated from personality, sexual arousal, and social perception variables along with self-reported likelihood to rape. In addition, zeroorder correlations also showed a significant relationship between sexual coercion and Rape Myth Acceptance, an attitudinal variable, and Psychoticism, a personality variable.
Factors Related to Self-Reported and Physiologically Measured Sexual Arousal at Rape Depictions
In an earlier study, Malamuth and Check (1983) showed that self-reported sexual arousal to rape depictions was predicted by Psychoticism and Neuroticism from the Eysenck Personality Inventory, self-reported likelihood to rape, power motivation, and sexual experience. Because some variables utilized in the present investigation were similar and because there may be different factors that predict self-reported or physiological arousal as compared to actual behavior, a second set of regression equations was performed. The Session 1 laboratory assessment in this study, which was most similar to Malamuth and Check's, was employed. All variables previously described, except sexual arousal and coercive behaviors, were included as predictor variables, and dependent variables were self-reported and physiological arousal to rape depictions. The results of these analyses are included in Table 5 for selfreported sexual arousal as the dependent variable and in Table 6 for physiological arousal as the dependent variable. In terms of the self-reported arousal, the overall multiple R was .44 (R2 = .19), which was significant (F = 4.81,d/ = 9, 185,p <.0001). Inspection of the beta weights indicated that Rape Myth Acceptance and the rape likelihood index were significantly related to the dependent variables, whereas the hostility discrimination index approached significance. Looking at the simple correlations, the rape likelihood index and the Rape Myth Acceptance scale were both highly correlated with the criterion, as was the hostility discrimination index, the Psychoticism scale of the Eysenck Personality Inventory and the Adversarial Sexual Beliefs scale.
For physiological sexual arousal, the multiple R was smaller (.36, R2 = .13), although still significant (F = 3.02, df = 9, 185,/» <.002). For this analysis, the rape likelihood index were significantly related to the dependent variables, whereas the only approached significance. In looking at the simple correlations, Rape Myth Acceptance and the rape likelihood index both were significantly correlated with the dependent variables, as was the Psychoticism scale of the Eysenck Personality Inventory. Results from this replication study were fairly similar to those of Malamuth and Check (1983), with the multiple Rs being somewhat smaller in this study.
The present investigation supports and extends previous literature (Abel et al., 1977; Burt, 1980; Giarrusso et al., 1979, Malamuth, 1981; Malamuth & Check, 1983; Rapaport & Burkhart, 1984) but raises a number of additional research issues. In interpreting the present results, it is important to inspect not only the beta weights but also the simple correlations between the predictors and the dependent variables. Because of correlation among predictors, certain variables that are related to the criterion may not receive significant beta weights because of this shared variance. It is important also to look at independent variables across the various measures of sexual aggressive behavior, i.e., self-reported coercive behavior, self-reported arousal to rape depictions, and erection responses to rape depictions.
In taking this view, certain variables are consistently correlated with the various dependent variables employed in this study; these include the self-reported likelihood to rape, Psychoticism on the Eysenck Personality Inventory, and Rape Myth Acceptance, replicating previous work with college student samples (Barnes et al., 1984a, 1984b; Malamuth, 1983; Malamuth & Check, 1980a, 1980b, 1983; Malamuth, Haber, & Feshbach, 1980; Malamuth, Heim, & Feshbach, 1980; Tieger, 1981). These results with a community sample again show that self-reported likelihood to rape contributes to the variance estimate of self-reported coercive behavior and arousal. This variable showed consistent relationship to the various measures of sexual coercion used in this study and seems to generalize well across samples. Similarly, Psychoticism and Rape Myth Acceptance were consistently related to the dependent variables and reinforce the view that individuals who engage in coercive sexual behavior have numerous misconceptions regarding rape, see women as wanting and desiring to be raped, and in general are aggressive and hostile individuals.
The measure developed specifically for this study, the hostility discrimination index, had the highest relationship with reported coercive sexual behavior and, in addition, related well to self-reported sexual arousal to rape depictions, although not showing a significant relationship to physiological measures. These findings indicate that subjects who show poor discrimination (i.e., smaller values) reported more sexual coercion as well as more sexual arousal to rape depictions. The results may be consistent with Malamuth and Check (1983): They have shown that measures of aggression and power, Psychoticism on the Eysenck Personality Inventory, and power motivation relate to reported arousal to rape depictions. In addition, as noted, feminist writers (e.g., Brownmiller, 1975) stress that rape is an aggressive rather than a sexual crime. The hostility discrimination index in this study did show a significant correlation with Psychoticism and Rape Myth Acceptance and may represent another measure of male hostile feelings toward women. However, the results of the present study did not allow clear determination of this, and the factors underlying the hostility discrimination index as defined in this study need further investigation with both normal samples and with arrested offenders.
The arousal measures, especially the rape index under arouse instructions, showed a marginally significant relationship with self-reported coercive behavior. Although the relationship was not as strong as in previous studies, it is consistent with data from clinical populations (Abel et al., 1977; Abel, Blanchard, Becker, & Djenderedjian, 1978; Barbareeetal., 1979; Quinsey & Chaplin, 1984; Quinsey etal., 1981) and college student populations (Check & Malamuth, 1983; Malamuth & Check, 1980b; Malamuth, Haber, & Feshbach, 1980; Malamuth, Heim, & Feshbach, 1980). However, sexual responding under suppress instructions failed to show a significant relationship with self-reported coercive sexual behavior. It may be that only in extremely coercive samples, such as clinically defined rapists, are such control mechanisms totally absent (Abel et al., 1980). In less coercive populations, it is possible that other factors lead to loss of control, such as anger or alcohol intoxication (Barbaree et al., 1983; Yates et al., 1984). Factors related to subjects' ability to control sexual arousal and situations under which such control mechanisms fail warrant further investigation.
The other variables investigated were somewhat more difficult to interpret or were inconsistent with the previous literature. Extraversion and Neuroticism both received significant beta weights in the multiple regression analysis for reported coercive behavior, but they showed no simple correlation with the criterion. These two variables also did not show any relationship to either self-reported arousal or physiological arousal, which suggests that Neuroticism and Extraversion, as measured in this study, related to coercive behavior as a result of the relation to other variables. Malamuth and Check (1983) have shown that Extraversion was highly correlated with sexual experience, which was highly related to arousal to rape depictions. In addition, Barnes etal. (1984a) have shown that individuals high in Extraversion have a rather hedonistic outlook toward sex and acquire sexual knowledge at a younger age. It may be that in the present study Extraversion relates to coercive sexual behavior because of its relationship to sexual experience. Individuals with greater sexual experience who are motivated by a hedonistic outlook toward sexuality may have more opportunities to be aggressive in sexual interactions. Also, Extraversion has been associated with antisocial characteristics, consistent with the findings of Rapaport and Burkhart (1984); therefore, in the present study it may be tapping that aspect of individual functioning. Further studies investigating the role of Extraversion in aggressive sexual behavior in terms of its covariance with antisocial characteristics or sexual experience in general are needed.
The role of Neuroticism in the current study is less clear. Malamuth and Check (1983) found that Neuroticism was correlated with self-reported sexual arousal to rape depictions, although Barnes et al. (1984a) found that the Neuroticism scale was not, in general, significantly correlated with various indices of sexuality employed in that study. On the other hand, Eysenck (1976) found that males high in Neuroticism were extremely conflicted about sexuality, having strong sexual desires but at the same time feeling very guilty regarding sexuality. Although such descriptions may be consistent with clinical descriptions of rapists (Groth, 1979; Rada, 1978), the role of sexual guilt and/or conflict needs further investigation in terms of its relationship to actually engaging in coercive behavior.
Sex Role Stereotyping and Adversarial Sexual Beliefs showed few correlations with any of the dependent variables employed in this study except for a zero-order correlation between Adversarial Sexual Beliefs and self-reported sexual arousal to rape stimuli. The results with Sex Role Stereotyping are inconsistent with previous work by Check and Malamuth (1983) and Koss et al. (1985) but are consistent with that of Rapaport and Burkhart (1984) and somewhat consistent with that of Ageton (1983) with an adolescent population. Rapaport and Burkhart did not find Sex Role Stereotyping related to actual sexual coercion in a college student population, and Ageton did not find the traditional sex roles related to actual aggressive sexual behavior in adolescents when delinquency behavior was controlled. Malamuth (1983) found that Sex Role Stereotyping is related to laboratory aggression against females in a college student population, although no measure of self-reported coercive sexual behavior was employed in that study; Koss et al. did employ a measure of coercive behavior. Results from the Adversarial Sexual Beliefs scale were also inconsistent with Rapaport and Burkhart, who found Adversarial Sexual Beliefs related to actual sexual coercion. Because of the difference in population-i.e., a community sample and a college student population-the actual roles of Adversarial Sexual Beliefs and Sex Role Stereotyping in sexual coercion await further replications. The overall results of the attitudinal measures suggest the need for further studies comparing the relative predictive value of more behavior-specific attitudes, such as Rape Myth Acceptance, versus more general attitudinal measures, such as Sex Role Stereotyping.
The major criticism that can be leveled against the present study is in terms of the definition of sexual coercion. A rather simple measure was employed, based on only three questions, and the possibility that there may have been individual subject variability in interpreting these questions cannot be ruled out. At the time this study was initiated (1981), more thorough measures of sexual coercion, such as those developed by Rapaport and Burkhart (1984) and Koss and Oros (1982) had not been published. However, when one looks at the results of this study in comparison to studies such as that of Rapaport and Burkhart, which used a more extensive sexual coercion history form, results are quite similar. In fact, the overall multiple Rs are almost identical to the multiple Rs in the Rapaport and Burkhart study, which was .45, whereas in the present study it was approximately .41.
In general, the results from this investigation indicate that variables drawn from both a sociocultural model and a psychopathology or individual difference model are related to various measures of aggression. As predicted, and consistent with previous literature, rape-supportive attitudes toward women, personality factors (especially those related to aggression, lack of empathy, and possible antisocial tendencies), and sexual arousal to rape depictions related at least to some extent to various estimates of sexual coercion. Males' perception of females also showed a relationship to sexual aggression and extends Abbey's (1982) earlier work. The present study also showed that results with a community sample are consistent with those found with college student samples. Factors not investigated in this study were situational factors, such as anger (Yates et al., 1984) and intoxication (Barbaree et al., 1983). To develop a more complete picture of the cause of sexual coercion against women, future investigations need to take into account these situational factors, in addition to factors drawn from the sociocultural and psychopathology models. The combination of these situational factors with such variables may provide a more complete picture of the cause of coercion against women and therefore assist in developing appropriate remediation and intervention programs.
1 The use of the lesbian film for the first stimulus was suggested by the reviewer of the grant that funded this project. The reviewer's belief was that the use of the lesbian film initially would "desensitize" subjects to laboratory procedures on a stimulus that was not to be used for data analysis.
2 The authors express appreciation to Dr. Neil Malamuth, who was a reviewer of the grant that supported this project, for suggesting the debriefing procedure.
3 A copy of the questionnaire is available from the first author.
Abbey, A. (1982). Sex differences in attributions for friendly behavior: Do males misperceive females' behavior? Journal of Personality and Social Psychology, 42, 830-838.
Abel, G. G., Barlow, D. H., Blanchard, E. B., & Guild, D. (1977). The components of rapists' sexual arousal. Archives of Sexual Psychiatry, 34, 895-903.
Abel, G. G., Becker, J. V., & Skinner, L. J. (1980). Aggressive behavior and sex. Psychiatric Clinics of North America, 3, 133-151.
Abel, G. G., Blanchard, E. B., & Becker, J. V. (1978). An integrated treatment program for rapists. In R. T. Rada (Ed.), Clinical aspects of the rapist (pp. 161-214). New York: Grune & Stratton.
Abel, G. G., Blanchard, E. B., Becker, J. V., & Djenderedjian, A. (1978). Differentiating sexual aggressives with penile measure. Criminal Justice and Behavior, 5, 315-332.
Ageton, S. S. (1983). Sexual assault among adolescents. Lexington, MA: D. C. Heath.
Albin, R. S. (1977). Psychological studies of rape. Signs: Journal of Women in Culture and Society, 3, 423- 435.
Armentrout, J. A., & Hauer, A. L. (1978). MMPIs of rapists of adults, rapists of children, and non-rapist sex offenders. Journal of Clinical Psychology, 34, 330-332.
Barbaree, H. E., Marshall, W. L., & Lanthier, R. D. (1979). Deviant sexual arousal in rapists. Behaviour Research and Therapy, 1 7, 252-259.
Barbaree, H. E., Marshall, W. L., Yates, E., & Lightfoot, L. O. (1983). Alcohol intoxication and deviant sexual arousal in male social drinkers. Behaviour Research and Therapy, 21, 365-373.
Barnes, G. E., Malamuth, N. M., & Check, J. V. P. (1984a). Personality and sexuality. Personality and Individual Differences, 5, 159-172.
Barnes, G. E., Malamuth, N. M., & Check, J. V. P. (1984b). Psychoticism and sexual arousal to rape depictions. Personality and Individual Differences, 5, 273-279.
Brownmiller, S. (1975). Against our will: Men, women and rape. New York: Simon & Schuster.
Burt, M. R. (1980). Cultural myths and supports for rape. Journal of Personality and Social Psychology, 38, 217- 230.
Check, J. V. P., & Malamuth, N. M. (1983). Sex role stereotyping and reactions to depictions of stranger versus acquaintance rape. Journal of Personality and Social Psychology, 45, 344-356.
Check, J. V. P., & MaJamuth, N. M. (1984). Can there be positive effects of participation in pornography experiments? The Journal of Sex Research, 20, 14-31.
Cooley, W. W., & Lohnes, P. R. (1971). Multivanate data analysis. New York: John Wiley.
Dixon, W. J. (1981). BMDP Statistical Software. Berkeley, CA: University of California Press.
Donnerstein, E., & Berkowitz, L. (1981). Victim reactions in aggressive erotic films as a factor in violence against women. Journal of Personality and Social Psychology, 41, 710-724.
Eysenck, H. J. (1976). Sex and personality. Austin, TX: University of Texas Press.
Eysenck, H. J., & Eysenck, S. B. G. (1975). Manual of the Eysenck Personality Questionnaire. San Diego, CA: EdITS.
Finkelhor, D. (1984). Child sexual abuse: New theory and research. New York: Free Press.
Giarrusso, R., Johnson, P., Goodchilds, J., & Zellman, G. (1979, April). Adolescents' cues and signals: Sex and assault. Paper presented at the meeting of the Western Psychological Association, San Diego, CA.
Groth, A. N. (1979). Men who rape: The psychology of the offender. New York: Plenum Press.
Kanin, E. J. (1969). Selected dyadic aspects of male sex aggression. Thf Journal of Sex Research, 5, 12-28.
Kanin, E. J., & Parcell, S. R. (1977). Sexual aggression: A second look at the offended female. A rchives of Sexual Behavior, 6, 67-76.
Kilpatrick, C., &: Kanin, E. (1957). Male sex aggression on a university campus. American Sociological Review, 22, 52-58.
Koss, M. P., Gidycz, A. C., & Wisniewski, N. (In press). The scope of rape: Incidence and prevalence of sexual aggression and victimization in a national sample of students in higher education. Journal of Consulting and Clinical Psychology.
Koss, M. P., Leonard, K. E., Beezley, D. A., & Oros, C. J. (1985). Nonstranger sexual aggression: A discriminant analysis of the psychological characteristics of undetected offenders. Sex Roles, 12, 981- 992.
Koss, M. P., & Oros, C. J. (1982). Sexual Experience Survey: A research instrument investigating sexual aggression and victimization. Journal of Consulting and Clinical Psychology, 50, 455-457.
Laws, D. R., & Osborn, C. A. (1983). How to build and operate a behavioral laboratory to evaluate and treat sexual deviates. In J. G. Greer & I. R. Stuart (Eds.), The sexual aggressor: Current perspectives on treatment (pp. 293-335). New York: Van Nostrand Reinhold.
Malamuth, N. M. (1981). Rape proclivity in the general public. Journal of Social Issues, 57, 130-157.
Malamuth, N. M. (1983). Factors associated with rape as predictors of laboratory aggression against women. Journal of Personality and Social Psychology, 45, 432-442.
Malamuth, N. M. (1984). Aggression against women: Cultural and individual causes. In N. M. Malamuth & E. Donnerstein (Eds.), Pornography and sexual aggression (pp. 19-52). Orlando, FL: Academic Press.
Malamuth, N. M. (1986). Predictors of naturalistic sexual aggression. Journal of Personality and Social Psychology, W, 953-962.
Malamuth, N. M., & Check, J. V. P. (1980a). Penile tumescence and perceptual responses to rape as a function of victim's perceived reactions. Journal of Applied Social Psychology, 10, 528-547.
Malamuth, N. M., & Check, J. V. P. (1980b). Sexual arousal to rape and consenting depictions: The importance of the woman's arousal. Journal of Abnormal Psychology, 89, 763-766.
Malamuth, N. M., & Check, J. V. P. (1983). Sexual arousal to rape depictions: Individual differences. Journal of Abnormal Psychology, 92, 55-67.
Malamuth, N. M., & Check, J. V. P. (1984). Debriefing effectiveness following exposure to pornographic rape depictions. The Journal of Sex Research, 20, 1-13.
Malamuth, N. M., & Donnerstein, E. (Eds.). (1984). Pornography and sexual aggression. Orlando, FL: Academic Press.
Malamuth, N. M., Haber, S., & Feshbach, S. (1980). Testing hypotheses regarding rape: Exposure to sexual violence, sex differences, and the "normality" of rapists. Journal of Research in Personality, 14, 121- 137.
Malamuth, N. M., Heim, M., & Feshbach, S. (1980). Sexual responsiveness of college students to rape depictions: Inhibitory and disinhibitory effects. Journal of Personality and Social Psychology, 38, 399- 408.
Murphy, W. D., Coleman, E. M., Haynes, M. R., & Flanagan, B. (1982, April). The measurement of social perception. Paper presented at the Fourth National Conference and Workshops on Sexual Aggression, Denver, CO.
Murphy, W. D., Krisak, J., Stalgaitis, S., & Anderson, E. (1984). The use of penile tumescence measures with incarcerated rapists: Further validity issues. Archives of Sexual Behavior, 13, 545-554.
Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner, K., & Bent, D. H. (1975). SPSS: Statistical package for the social sciences. New York: McGraw-Hill.
Panton, J. H . (1978). Personality differences appearing between rapists of adults, rapists of children and non-violent sexual molesters of female children. Research Communication in Psychology, Psychiatry, and Behavior, 3, 385-393.
Quinsey, V. L., Arnold, L. S., &: Pruesse, M. G. (1980). MMPI profiles of men referred for a pretrial psychiatric assessment as a function of offense type. Journal of Clinical Psychology, 36, 410-417.
Quinsey, V. L., & Chaplin, T. C. (1984). Stimulus control of rapists' and non-sex offenders' sexual arousal. Behavioral Assessment, 6, 169-176.
Quinsey, V. L., Chaplin, T. C., & Upfold, D. (1984). Sexual arousal to nonsexual violence and sadomasochistic themes among rapists and non-sex-offenders. Journal of Consulting and Clinical Psychology, 52, 651-657.
Quinsey, V. L., Chaplin, T. C., & Varney, G. (1981). A comparison of rapists and non-sex offenders' sexual preferences for mutually consenting sex, rape, and physical abuse of women. Behavioral Assessment, 3, 127-135.
Rada, R. T. (1978). Clinical aspects of the rapist. New York: Grune & Stratton.
Rapaport, K., & Burkhart, B. R. (1984). Personality and attitudinal characteristics of sexually coercive college males. Journal of Abnormal Psychology, 93, 216-221.
Russell, D. E. H. (1982). The prevalence and incidence of forcible rape and attempted rape of females. Victimology: An International Journal, 7, 81-93.
Spence,J., Helmreich, R., & Stapp,J. (1973). A short version of the Attitudes toward Women Scale (AWS). Bulletin of the Psychonomic Society, 2, 219-220.
Tieger, T. (1981). Self-rated likelihood of raping and the social perception of rape. Journal of Research in Personality, 15, 147-158.
Yates, E., Barbaree, H. E., & Marshall, W. L. (1984). Anger and deviant sexual arousal. Behavior Therapy, 15, 287-294.
Zuckerman, M. (1971). Physiological measures of sexual arousal in the human. Psychological Bulletin, 75, 297-329.
Acknowledgments. Portions of this project were supported by USPHS Grant MH-34030-02 from the Center for the Studies of Crime and Delinquency and from Tennessee Department of Mental Health and Mental Retardation Research Contract #ID-0783 to the Department of Psychiatry. The authors wish to express appreciation to Donna Townsend and Bette Ackerman for secretarial and editorial assistance.
Reprints: Requests for reprints should be directed to William D. Murphy, Department of Psychiatry, University of Tennessee, Memphis, Tennessee 38105.
William D. Murphy*
Emily M. Coleman[dagger]
Mary R. Haynes[double dagger]
*Department of Psychiatry, University of Tennessee, Memphis-The Health Science Center.
[dagger]Forensic Psychiatry Clinic, Bronx-Lebanon Hospital Center, Bronx, New York.
[double dagger]Department of Psychiatry, University of Tennessee, Memphis-The Health Science Center.…