Rapists\' Offense Processes: A Preliminary Descriptive Model

Polaschek etOF JOURNAL al. INTERPERSONAL / RAPISTS’ OFFENSE VIOLENCE PROCESSES / June 2001 Models of the offense process set out to provide a description of the cognitive, behavioral, motivational, and contextual factors associated with a particular type of offense. They model the temporal relationships between the variables of interest and focus explicitly on the proximal causes or the how of offending. In this study, qualitative analysis was used to develop a descriptive model of the offense process for 24 adult rapists. Sentenced offenders provided detailed retrospective descriptions of their thoughts, emotions, and behavior prior to and during their most recent rape. These descriptions were analyzed using a grounded theory approach. The resulting preliminary model contained the following six phases: background factors to the offense, goal formation, approach behavior, offense preparation, the offense, and postoffense behavior. These stages are discussed along with the model’s theoretical, research, and clinical implications. The model is consistent with existing multivariate theories of rape and has clear taxonomic potential.

Rapists’ Offense Processes A Preliminary Descriptive Model DEVON L. L. POLASCHEK Victoria University of Wellington

STEPHEN M. HUDSON University of Canterbury

TONY WARD University of Melbourne

RICHARD J. SIEGERT Victoria University of Wellington

The application of relapse prevention (RP) theory to the treatment of sexual offenders has had many beneficial effects. There has been remarkable growth in its use in sex offender treatment programs (Marshall, Fernandez, Hudson, & Ward, 1998). In addition, the model has been developed and further adapted (Laws, 1989, 1995; Pithers, 1990). The RP model recognized that relapse into addictive behavior does not occur unexpectedly but is the result of a chain of internal and external events that are potentially identifiable for each individual. The important contributions of the RP model were (a) the development of more elaborate self-management strategies to enable clients to maintain behavioral change initiated in treatment and (b) the introduction of the central notion that relapse was part of an unfolding sequence of behavior over time. In its application to criminal offending, this sequence has come to be called the offense chain or offense process. In this article, we develop a JOURNAL OF INTERPERSONAL VIOLENCE, Vol. 16 No. 6, June 2001 523-544 © 2001 Sage Publications, Inc.

523

524

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

tentative model of the offense process for rapists based on offenders’ descriptions of their own offenses. Research investigating the validity of the RP model as a template for different forms of offending has been sparse. In the mid-1990s, our research suggested a number of theoretical and clinical problems in the way in which it had been adapted for use with child sex offenders (Ward, Hudson, & Marshall, 1994; Ward, Hudson, & Siegert, 1995). One of the implications of these findings was that it would be worth building descriptive models of the offense process directly from data provided by offenders about how their assaults unfolded, enabling examination of the similarities and differences between the processes offenders actually reported and what the RP model suggested. The first descriptive model constructed from offender data was of child sexual offending. Ward, Louden, Hudson, and Marshall (1995) applied qualitative data analysis techniques to offense processes from 26 incarcerated child sex offenders. Their final model contained nine stages describing the cognitive, affective, and behavioral sequence of events. More important, they found frequently followed pathways that were not captured by the RP model (Hudson, Ward, & McCormack, 1999). For instance, the RP model describes offending as occurring when an internally generated prohibition against the offensive behavior is violated. This loss of restraint is facilitated by a build-up of stress, resulting in negative affect. The model drew attention to offenses that were committed primarily in states of positive affect, that were not triggered by building stress, and in which the offender had acquisitive or approach rather than avoidance goals (Ward, Hudson, & Keenan, 1998). The Ward, Louden, et al. (1995) descriptive model elaborated on the mechanisms that provided transitions from one stage of the offense process to another. In particular, it delineated the various roles that cognitive distortions, sexual arousal, and affective states played at different points in the process of offending. Overall, the following three predominant patterns described the course each offense charted through the model: a positive affect pathway with explicit planning and cognitions supportive of adult-child sex; a negative affect pathway characterized by implicit planning, offender neediness, feelings of entitlement with regard to sex, and negative evaluations of offending behavior; and finally, a pattern of shifting between the first two (Hudson & Ward, 1996). However, there were a number of limitations. Although their model accommodated offender heterogeneity, it was still limited by its reliance on data from offenders who had volunteered for treatment. It needs a broader sample to increase heterogeneity further. The low level of detail provided by offenders in some areas gave rise to other limitations. For example, it permitted little analysis of offender behavior during the offense

Polaschek et al. / RAPISTS’ OFFENSE PROCESSES

525

itself and led to a lack of clarity about whether victim behavior had any influence on offenders. We wanted to address these limitations in the research presented here. Those who work with men who have committed sexual offenses observe that there are similarities and differences between those who sexually abuse children and those who rape adults (Pithers, 1993). An encouraging number of broad, multivariate etiological theories have been proposed for rape. However, at the more specific levels of theorizing, rapists have often been collapsed together with child sex offenders, and the nature of the actual mechanisms involved in their offending have been poorly specified (Polaschek, Ward, & Hudson, 1997). Moreover, empirical research has not yet established that standard cognitive-behavioral treatments for rapists are effective. The absence of research at the micro-model level has hampered the development both of higher level theory and classification and treatment programs that better capture the dynamic risk factors of rapists. Our recent research with child sex offenders has been fruitful in these domains, and this strongly suggests that a parallel endeavor with rapists has the potential to yield a variety of interesting and useful results. Models of the offense process serve to both ground the provision of treatment and the development of higher order theory. Therefore, the purpose of this study was to develop a preliminary descriptive model of the offense process for men who had sexually assaulted adults.

METHOD

Participants The research participants were 24 men currently serving a prison sentence for sexual violation or attempted sexual violation (i.e., rape) of a person older than the age of 16 years. They were from five prisons throughout New Zealand. Of the first 32 participants who were approached, 8 declined, yielding a participation rate of 75%. The mean age, sentence length, number of previous convictions, and mean age at first conviction of participants and those who declined to participate did not differ significantly. The mean age of participants at the time of the index sexual violation was 31.8 years (SD = 7.6, range = 21 to 44 years). Mean sentence length was 7 years (SD = 2.3, range = 3 to 10 years). Two participants were on indefinite sentences. Mean age at first conviction was 23.0 years (SD = 8.5, range = 15 to 45). Five had no previous convictions. The remaining 19 participants had a

526

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

mean of 13.3 previous convictions, most of which were for property, driving, and minor drug offenses. Ten had previous convictions for violence and 5 for sexual offending. The offenses for which participants were serving their current sentences (the index offenses) included 31 sexual violations involving 27 victims (i.e., some offenders were convicted on two or more charges against the same victim). One victim was a male. The victims of 7 offenses were current or estranged wives or cohabiting partners. Nine victims were strangers, and 8 were casual acquaintances. Four offenses were committed by fathers and 3 by close friends. Approximately half of the men admitted their offending, but 11 denied the offenses in that they asserted that they had been involved in consensual sexual activity with the victim. Five men had received psychological treatment related to their offending since commencing their sentence. Procedure All data were collected by the first author. Demographic and preliminary offense-related information were obtained from participants’ institutional files. These files varied in the amount of independent information they contained that could be used for verifying offenders’ accounts of their offensive behavior. Following perusal of the file information, interviews were conducted with participants focusing on the index offense and events leading up to it. Interviews were audio taped for later reference. However, we obtained the data presented here mainly from the offender’s own oral narrative of the offense, collected from him during the interview. Interview procedures for obtaining the oral narrative incorporated a number of techniques used routinely in scientific clinical practice to enhance the reliability and validity of the data obtained (see Zamble & Quinsey, 1997). In inquiring about the offense process, the interviewer began with open-ended questions, gradually funneling in to establish fine details. Intentional redundancy was built into inquiries; the interviewer would return to key areas with questions in different forms to assess accuracy, consistency, and the level of detail provided. Where the offender’s account differed from that of the file information, the discrepancy was drawn to his attention along with a nonjudgmental request for its resolution. The offender’s responses to these requests were included in the data for later analysis. The interviewer typed the resulting narrative into a portable computer during the interview to form the offense process. Most participants were able to describe the events that unfolded relatively readily but required prompting from the interviewer to include details of their thinking, affective responses,

Polaschek et al. / RAPISTS’ OFFENSE PROCESSES

527

and the volitional components of the resulting offense process. The section of the interview in which data for the offense process were gathered generally took from 1 to 3 hours to complete.

RESULTS

Data Analysis Procedure Once the offense processes were constructed, we analyzed them using Strauss and Corbin’s (1990) grounded theory procedures. In the first stage, the first author broke down the narrative vignettes into meaning units, small blocks of text containing single phrases or sentences. Next, we coded the meaning units into provisional categories on the basis of semantic similarity with other units. Each category was assigned a descriptive name that best captured the common semantic content of the category at the lowest possible level of abstraction. Each meaning unit was assigned to one or more categories (multiple coding), and a new category was created if the concept did not fit any existing category. We used these steps to code the first 6 offense descriptions. When this coding was completed, provisional categories were combined or subsumed under other, more abstract categories. Next, we coded further meaning units from new offense-process-based interviews and used these to test the adequacy of the provisional categories from the previous stage. Individual categories were expanded, or two or more categories were collapsed together (i.e., combined and given a more abstract label) to accommodate more data, and where necessary, new categories were developed to allow new data to be incorporated. When 10 vignettes had been coded, we used the derived categories to develop a preliminary model of the temporal relationships between categories. The process of checking back through the vignettes as the model was developed resulted in further collapsing and refinement of categories. Finally, we coded the remaining 14 offense descriptions, categorized them, and used the results to check the accuracy of the model. Yet more refinement and abstraction as well as the addition of some new categories led to the final model. We constructed a diagrammatic representation containing 21 categories and 43 subcategories. Two or more investigators worked together on all phases of data analysis beyond the initial coding of meaning units to reduce the influence of individual biases in model development.

528

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

Background factors Generally positive

Generally negative Management (coping style) Emotion-based

Problem-based Proximal mood

Positive

Depressed

Angry

Establishment of dominant goals

Figure 1:

Phase I: Background

For clarity of presentation, the model is divided into six phases containing between three and five main categories each. The six phases were background, goal formation, approach, preparation, offense, and postoffense. The Descriptive Model Next, we describe the model phase by phase. Italics in the text refer to actual category or subcategory labels in the corresponding figures. Parentheses contain examples taken from the narratives of individual participants. Phase I: Background The first major category of the model, background factors, covers the offender’s lifestyle and circumstances in the days, weeks, or months prior to offending (see Figure 1). It includes employment, relationships with others, leisure time, finances, and alcohol and drug use as well as the effect of any historical difficulties on offenders’ day-to-day living. The two subcategories reflect the overall affective influence of this background on the offender— generally positive (on the left of Figure 1) (“I have a direction now, I know what I want to do. Things are stable and positive.”) and generally negative (on the right side of Figure 1) (“I’ve been feeling pretty down during this time, just feeling lonely and lost.”). The next category, management (coping style), contains two subcategories. The first, on the left-hand side of the model, is

Polaschek et al. / RAPISTS’ OFFENSE PROCESSES

529

problem-based, meaning that problems that arose were addressed effectively, with the offender often regaining positive affect as a result (“She didn’t want to live with me any more but I’ve managed to reconcile with her, everything is coming right again.”). On the right-hand side is emotion-based management. Here, the offender attempted only to ameliorate the resulting negative affect rather than directly focusing on solving the problem. An example of this would be when an offender’s response to a deteriorating intimate relationship is to consume more alcohol, have sex with prostitutes more often, and smoke more marijuana. Over several months, these coping strategies create financial stress. Also, they result in more discord with his girlfriend and more pressure from his employer because his work performance is negatively affected. Emotion-based coping often resulted in temporary positive affect, but in the long-term, stress and negative affect escalated because of the accumulation of unsolved problems. The last major category of the first phase is proximal mood, which refers to events of the few hours leading up to the offense. There were three subcategories, positive, depressed, or angry. Proximal mood and the events leading to it were influential in determining the next category, establishment of dominant goals. Phase II: Goal Formation Phase II (see Figure 2) begins with the process of establishment of dominant goals that direct the subsequent behavior and may ultimately guide the commission of the offense. These goals were labeled dominant to recognize that offenders often described a number of goals that were subservient to them. At this early stage of the offense process, two dominant goals emerged from the offender’s proximal mood state. The first was seeking sexual gratification (“We hang around the pub for a while. She [an attractive stranger] is there. She looks [attractive]. I wonder if I’ll get an opportunity to have sex with her tonight.”). Those seeking sexual gratification were doing so either to enhance a preexisting positive mood (e.g., after an enjoyable party finished early, one offender said “[The decision to have sex] just seems to be a good idea to fulfill the whole evening.”) or to escape a preexisting negative mood (“I’m kissing and cuddling this other girl hoping it might lead to something more. . . . Maybe I’m already thinking, ‘I’ve blown it.’”). The second type of dominant goal was labeled redressing harm to self because this goal was set by men who perceived that a particular woman’s behavior had hurt them and who intended to seek some form of redress of that harm. There were important differences in how men initially sought to redress this harm. For instance, some men chose victim harm: They immedi-

530

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

Proximal mood Establishment of dominant goals

Seeking sexual gratification

Redressing harm to self

Enhancement of

Escape from

positive mood

negative mood

Victim harm

Interpersonal problem-solving

Distal goal-related planning

Planning sexual access

Sexual assault

Planning to redress harm (non-sexually)

planning

Encounter with victim

Figure 2:

Phase II: Goal Formation

ately sought redress by deciding to harm someone. Usually, the targets were the women perceived to have harmed them, but occasionally, their harmful intent had a general focus (“I blamed women instead of myself, I wanted to get some of my self back, my self-esteem, by making them feel useless.”). Still others initially wanted to sort out the problem in a prosocial manner (interpersonal problem solving) (“I want to try and clear the thing up. . . . I’m just getting humiliated every time I see her. I’m hoping for [an apology].”) Once they had established goals, offenders began planning. Their explicit plans had one of three foci. The first, on the left-hand side of Figure 2, was how to gain sexual access (“Everyone else is still out, and it just sort of occurs to me ‘Oh, I’ll just go in and try my luck.’”). Here, the offender was planning to have sex but not planning to use force or proceed without the consent of the targeted woman. The second planning focus was on how to commit a sexual assault (“So I thought, ‘I’ll have sex instead, I’ll go and rape someone.’”). Finally, on the right side of the model, men were planning to redress harm. Note that harm redress plans did not yet contain sexual components. However, they often contained criminal elements such as the intent to physically assault the victim (“Next one in is gonna get a smack in the head.”). A few offenders appeared to do no planning at all, moving straight from deciding on their goal to enacting it.

Polaschek et al. / RAPISTS’ OFFENSE PROCESSES

531

Distal goal-related planning

Encounter with victim Communication of intent Indirect

Direct Victim response

Compliant

Resistant

Offender evaluation of progress toward goal Positive

Negative Establishment of secondary goal

Sexual gratification

Victim harm

Proximal approach strategy selection Decision to commence sexual activity

Figure 3:

Phase III: Approach

Phase III: Approach The approach phase (Figure 3) begins with an encounter with the victim, initial approach behavior that leads to the communication of intent to the victim. Communicating intent was indirect for offenders who were grooming a victim to gain consent for sex rather than engendering resistance. These offenders were engaging in a process that was colloquially referred to as “chatting her up.” Generally, the offender avoided any verbal interaction with the victim that was explicitly about consent (“I’m chatting her up to see if there’s a chance. . . . I’m trying to steer the conversation as best I can.”). With direct communication, the offender informed the victim directly of what his goal was. Either he did so verbally or he used direct physical force, abducting the victim or threatening her with violence. Offenders reported victims’ responses to this communication as either compliant or resistant. Victims’ behavior was reported as more actively resistant in the approach phase than during the assault itself. However, offenders also reported compliant behavior. They often interpreted apparent compliance as implicit consent.

532

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

An offender’s evaluation of progress toward his goal was usually based on his interpretation of the victim’s responses. Offenders made positive evaluations when they perceived either that the victim was consenting to sex or that they had her under their control (“She starts to get undressed, she’s obviously conceded.”). In negative evaluations, the victim was perceived to be thwarting the offender or disrupting his goal achievement (“She doesn’t take me serious. . . . She thinks it’s all a bit of a joke really.”). Following evaluation, about one third of the sample went through the establishment of a secondary goal. Secondary goals reflected a change in direction stimulated by the preceding evaluation of progress. Rapists either added an extra goal to their original one (e.g., obtaining sexual gratification as well as harming her) or changed the original goal for a new one (e.g., deciding now on victim harm instead of resolving interpersonal conflict). It is interesting that men who had just made negative evaluations were most likely to form secondary goals. Some men who originally sought harm redress now formed secondary sexual gratification goals, whereas others developed secondary victim harm goals. Some men who initially had sought sexual gratification now also formed victim harm goals. Men who now became intent on victim harm were often angry after evaluating their progress because they considered that the victim was not cooperating with their plans. The length of the approach phase varied greatly. Some offenders repeated the approach sequence more than once; their behavior is represented by the recursive arrows on the left side of Figure 3. The repetition was either directed at one potential victim or a series of possible victims. However, most men, especially those who evaluated their progress as positive, decided to commence sexual activity in the first iteration and so went straight on into the preparation phase. Phase IV: Preparation Preparation for the sexual offense commenced with the decision to commence sexual activity (see Figure 4). Having made this decision, a few offenders immediately began their sexual assault. However, most first appraised the expressive potential of the situation. This appraisal referred to whether they judged that they could express themselves unconstrained by environmental or practical factors or whether instead, they judged their behavior to be constrained by instrumental considerations. Constraints came mainly from concern about victim control and the risk of detection by others (“I don’t want to take my clothes off and leave them around, try to find them. . . . I don’t know where her boyfriend is so I gotta get out of here.”). Men who were concerned about constraints often committed their rapes very

Polaschek et al. / RAPISTS’ OFFENSE PROCESSES

533

Proximal approach strategy selection Decision to commence sexual activity Appraisal of expressive potential of situation Constrained by instrumental considerations

Unconstrained Preparation for sexual assault Circumventing victim resistance

Arousing self

SEXUALLY ASSAULTATIVE BEHAVIOUR

Figure 4:

Phase IV: Preparation

quickly or interrupted their sexual activity periodically to reassess constraining factors or to exert more control over the victim through renewed violence or threats. Having appraised the expressive limitations of the circumstances, many offenders engaged in sexual preparation. Here, they either did things intended to enhance their own level of arousal (“I make her handle me so I get aroused enough to rape her.”), or they engaged in behavior usually associated with conventional foreplay. They appeared to do this second form of preparatory behavior either for the purpose of maintaining the illusion of victim consent or in an effort to arouse the victim and so undermine her resistance (“I’m pushing her towards the bedroom. She falls back on the bed and I fall on top of her. I start playing, pulling her pants down, she’s pulling them back on, we’re play fighting.”). The preparation phase was often of short duration. Phase V: Offense Offenders reported a range of sexual behavior during the offense itself (see Figure 5). The most common sexual behavior was conventional penilevaginal intercourse: nongratuitously degrading sexual behavior. Gratuitously degrading sexual behavior was less common; it included anal intercourse, penetration of the victim’s vagina with objects, and forcing the victim to perform fellatio. Whether something was classified as gratuitously degrading or not depended entirely on whether the offender thought or intended that

534

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

Preparation for sexual assault

SEXUALLY ASSAULTATIVE BEHAVIOR

Non-gratuitously degrading

Gratuitously degrading

sexual behavior

sexual behavior

Gratuitous physical violence Victim response during assault

Compliant

Resistant

Offender evaluation

Positive

Negative

Immediate post-rape situation management

Figure 5:

Phase V: Offense

it was. However, the two categories did not overlap because in this sample, offenders mainly viewed behavior other than conventional intercourse as degrading. Moreover, those who intended to degrade were much more attentive to the victim’s responses than those who were focused on their own sexual enjoyment (“Groping her genitals, I want to see the reaction in her face. I want it to piss her off.”). Gratuitous physical violence was quite common in the approach phase, but during the sexual assault, it occurred rarely and only when offenders became frustrated with the victim. Otherwise, violence was limited to that needed to control the victim. Gratuitous violence only occurred in conjunction with gratuitously degrading sexual behavior. This is reflected in its placement on the right-hand side of the model. Again, offenders reported the victim’s response during the assault either as resistant or compliant. It was followed by an offender evaluation that was either positive or negative. Offenders reported affective responses consistent with their evaluations. At this stage, they tended to evaluate whether they had achieved their dominant or secondary goals, such as harming or punishing the victim or “getting laid.” Offenders with a focus on their own gratification often reported both positive evaluations and affect. However, there were notable exceptions: They were negative if the victim’s behavior had been sufficiently resistant to interfere with achieving what they wanted, and some of these offenders were

Polaschek et al. / RAPISTS’ OFFENSE PROCESSES

535

disappointed because having sex with the victim did not live up to preconceived fantasy (“I thought it would’ve been a bit more enjoyable, but it was just a matter of do it and get it over with.”). Those whose goals were victim harm oriented were either satisfied (“I got my revenge anyway.”) or started to appraise the enormity of what they had done even before the offense was completed (“I don’t get a buzz or anything out of it at any point. I don’t feel any major pleasure except for the tenth of a second when I’m ejaculating. I know I’m a goner.”). Many of the sexual offense phases were brief. However, several lasted 1 or more hours in which there were periods of sex interspersed with nonsexual activity. In these cases, the offense followed the recursive pathway from offender evaluation back up to commencing sexually assaultative behavior. As in the previous phases with recursive arrows, in repetition, such offenses did not necessarily follow exactly the same pathway back through the subcategories. For instance, an offender who intended to degrade the victim reported his victim’s behavior to be resistant in the first iteration. He evaluated this positively—he was achieving his intention of humiliating her. He continued his offensive behavior, but subsequently, the victim’s behavior altered to more passive compliance. The offender became dissatisfied and ceased the assault. Phase VI: Postoffense Following the completion of the sexual assault, perpetrators immediately began to take stock of the consequences. We called this phase immediate (postrape) situation management (see Figure 6). Offenders became concerned with managing the victim and the environment. Some responded to this concern with normalizing behavior, the function of which seemed to be to maintain their own or the victim’s perception that they had just been involved in consenting sex (e.g., cuddling the victim, offering her a cigarette, and talking to her about her work). Others used overt control strategies to control the immediate risk of apprehension (e.g., tying her up, cutting phone lines, and escaping). Some offenders did a mixture of both classes of behavior. In the next part of this phase, rapists conducted an evaluation of the situation. This evaluation was broader than that at the end of the previous phase. It included more factors and both a review of the past and consideration of future issues. Evaluation of external factors included victim effect and ongoing risk of detection (“Then I start analyzing what I’ve done, what I’ve actually done to her. What am I going to do now to repair what I’ve done?”). Self-evaluation was of whether dominant goals were achieved (“I don’t give a

536

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

Offender evaluation Immediate (post-rape) situation management

Normalizing behavior with victim

Overt control strategies

Evaluation of situation

Self

External factors

Affective response

Positive/neutral

Angry

Other negative

Behavioral response

Yes

No

Long-term post-offense responses

Figure 6:

Phase VI: Postoffense

damn about whether I get in trouble. I keep thinking that’s my revenge.”) and both the short- and long-term implications of the interaction (“I’m realizing what I’ve done and I don’t know whether to go straight to the police.”). Some offenders engaged in both forms of evaluation. This evaluative phase generated a range of affective responses. At this point, some offenders suddenly experienced guilt, shock, or fear (other negative) as they realized for the first time that they had committed a serious crime (“What I’m mostly feeling now is fear. I’m realizing what I’ve done and I don’t know whether to go straight to the police.”). Others became angry. Some felt positive/neutral (“By now I’m drained, it’s been a long night. A bit contented.”). In response to these evaluations and affective responses, some offenders selected and implemented a behavioral response, a purposive response to their evaluation such as attempting to conceal the crime or going into hiding (“I’m thinking about my own safety now. I head for my car and drive quite a way into the bush.”). However for many, including most of those who had persuaded themselves that their behavior was not criminal, they simply went

Polaschek et al. / RAPISTS’ OFFENSE PROCESSES

537

home or went to sleep. Intriguingly, most of the offenders in this sample made no serious attempt to evade arrest even when they acknowledged the criminality of what they had done. This oversight no doubt accounts in part for how they came to be incarcerated. Finally, long-term postoffense responses included the offender’s choice to seek help in recognition that he had a problem, continue an existing criminal lifestyle, or develop strategies for avoiding future apprehension.

DISCUSSION

This preliminary rape model (RM) demonstrates the utility of a grounded theory approach in analyzing offenders’ narratives of their offense processes. The approach was successful in forming individual offense descriptions into a cohesive model that reduces the data but still captures both the richness and the heterogeneity of the original information. Although preliminary, the resulting model has a number of interesting features. We will discuss these first and then consider the theoretical and clinical implications. Finally, we will consider some limitations of the study and future research directions. Descriptive models represent the most fundamental level of theory building (Ward & Hudson, 1998); they are judged successful to the extent that they capture the essential elements of how offenders go about committing their criminal acts. The RM elements are built from the interactions among cognitive, behavioral, affective, and volitional components. Few previous descriptive models have given much specific attention to rape’s volitional aspects despite decades of debate in the literature about the volitional underpinnings of the crime. One reason for this omission may be that RP-type models assume that the offender’s dominant goal throughout the offense process is to relapse (i.e., to offend sexually) and thus that all behavior is in the service of this goal (e.g., Carich, 1994; Freeman-Longo & Pithers, 1992). Of course, this assumption leads to the further assumption that from early in the offense process, all offenders are planning to commit a sexual assault. Thus, where no overt evidence of sexual assault planning is uncovered, implicit or covert planning must also be assumed. Another reason that volitional aspects have been ignored could be that most descriptive models have focused primarily on sexual offending against children; the main goals suggested for child sexual abuse are sexual or intimacy related (Knight & Prentky, 1990). By contrast, we constructed a model around offenders’ explicit goals and found that the goal of committing a sexual offense was only rarely explicit early in the offense process. Using explicit rather than inferred goals as an organizing tool was a more rationally defensible strategy in this research for

538

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

two reasons. First, a wider range of goals have been implicated in sexual aggression toward adults, including sex, antisocial opportunism, intimacy, and victim harm goals (see Knight, Rosenberg, & Schneider, 1985; Pithers, 1993). Second, in our sample there was little evidence of previous experience with sexual aggression, either from self-report or file information. Neither did men generally report any forewarning of where the interaction with the victim might be headed. Moreover, there are many legitimate reasons for men to spend time alone with adult women. Considering all of these points together, we did not have any evidence that sexual goals were more likely to be operating implicitly than at any other time in the offenders’ lives. Taking goals at face value avoids the problem of suggesting that we as researchers have privileged access to offenders’ mental processes. Put another way, the RM is an offense model. It is not a relapse model, although it would accommodate a classical relapse process. We have argued that relapse and etiological models are similar (Hudson & Ward, 1996) in the sense that an offense model is broader and should be able to subsume a classic relapse pathway. However, it is notable that in the offense processes on which the RM was built, there was little evidence of the classic relapse pathway. An important finding of the approach we have taken to modeling rape is that one of the significant sources of heterogeneity in rape comes from the disparate range of distal goals. These goals have taxonomic implications that have often been recognized in rape classification systems. For example, Groth, Burgess, and Holmstrom (1977) regarded anger and power as the essential motives that distinguish rapes from each other. Similarly, Knight and Prentky’s (1990) empirical classification system (the Massachusetts Treatment Center: Rape Version 3) used four primary motivations—opportunistic, pervasively angry, sexual, and vindictive. These existing taxonomies have treated offender goals as if they are temporally stable throughout the offense process. The RM demonstrates that such stability cannot and ought not be assumed for some offenses. Particularly where initial goals were not sexual, significant goal shifts occur as the interaction unfolds. So, an offender can go from having a sexual goal to revenge and from a goal that is not criminal at all, such as persuading the victim to apologize, to intentional victim harm. This finding is consistent with our recent reconceptualization of offense process models within a goal theory framework (Ward et al., 1998). The incorporation of these goals shifts attention to another strength of the RM—the focus on interactions between the offender and the victim. Rape— indeed all violent offenses—can be viewed as a particularly offensive form of social interaction (Canter, 1994). This social context and especially the role of victim behavior in rape have largely been ignored in research; the offender’s

Polaschek et al. / RAPISTS’ OFFENSE PROCESSES

539

criminal acts seem often to be presented as if they were carried out in a socially static context. Several studies have demonstrated that victim behavior does have an effect on how the offender behaves and on whether he even completes the offense (e.g., Ullman & Knight, 1995). However, these studies were interested in the effects of different forms of victim resistance; they are not specific about the points at which victim behavior is most noted by offenders, and neither do they code victim behavior in terms of how the offender interpreted it. Although the RM is based mainly on offenders’ accounts of victims’ behavior, which no doubt incorporate self-serving distortions, the RM demonstrates that victim behavior does affect how the offense unfolds and may even affect whether a rape occurs (instead of some other type of offense), especially on the right-hand side of the model, when the early plans are not sexual in nature. The RM also demonstrates the key points at which offenders were particularly aware of victim behavior in each phase after victim contact is established and how their perceptions of victims altered both their goals and the course of their subsequent actions. The RM is generally consistent with existing multivariate theories of rape. Men who predominantly followed the left-hand side of the model (positive affect throughout the offense, low violence offenses, common denial of the intent to aggress, and difficulty accepting retrospectively that they did not have the consent of the victim) were well described by Marshall and Barbaree’s (1990) theory in which sexual offenders have failed to differentiate sexual and aggressive behavior in sexual contexts. These perpetrators’ interpersonal offense style was also well captured in Malamuth’s (Malamuth, Heavey, & Linz, 1996) confluence model in which men are said to develop a view of sex as an impersonal utility. Malamuth et al.’s (1996) unconvicted college student sample may provide a particularly good comparison with the sample used in this study as these men generally had no previous convictions for rape and had committed offenses that were relatively low in violence. Significantly, however, salient elements of offense patterns that follow a more right-handed course through the model are not particularly well captured by existing rape theories. The use of rape as a punishment or as a device to humiliate and degrade is implicit or is mentioned in passing by a number of theorists (Malamuth et al., 1996; Marshall & Barbaree, 1990; also see Darke, 1990); the significance of this pathway in the RM suggests that there is a need to give more emphasis to the mechanisms involved in these rapes. The pattern of addressing threats to self by either inflicting violence on the victim from the outset or attempting a more prosocial method of grievance resolution and then inflicting harm if that fails is not explicitly addressed. At the abstract level, this process more closely resembles that of a nonsexual

540

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

assault. Often, offenders on this path were indeed physically assaulting their victims during the approach phase. The main point of difference with nonsexual physical assault alone is that somewhere very proximal to the commencement of the sexual assault phase, the offender decides to add sexual assault elements to his interaction with the victim. Existing theories of violent offending may be a more appropriate future comparison for these nonsexual aspects of the model. Of potential relevance to the harm redress/victim harm offenses is Baumeister’s (Baumeister, Smart, & Boden, 1996; Bushman & Baumeister, 1998) recent research on the role of unstable high self-esteem (i.e., narcissism) in the use of aggression toward others. We did not measure narcissism in our research, but it may provide a theoretical mechanism for explaining why some men solve particular interpersonal problems in extreme and violent ways rather than simply walking away or negotiating a genuine compromise with the other party. The RM has some interesting implications for the treatment of rapists, an endeavor that currently leaves room for considerable improvement (Polaschek et al., 1997). Differing goals early in the offense process have clear implications for the situations in which particular men are most at risk. The two sexual gratification goals suggest that men are at risk who are seeking sex, often relatively impersonal sex, to augment a good mood or to escape low mood. These men are likely to commit low violence offenses in circumstances where issues of consent are intentionally unclear. They are relatively disinterested in victims’ responses and are insensitive to their cues. Clearly, such offenders are likely to be unskilled in the development and maintenance of intimate relationships. The goal of redressing harm has implications too for both treatment and posttreatment situational risk management. In the RM sample, the establishment and criminal fulfillment of a harm redress goal suggests a vulnerability to self-regulatory failure in response to negative evaluation by others. If an individual commits an offense as a consequence of such self-regulatory failure, then assessment should be undertaken to establish whether there is other evidence of such failures both in this specific context and in other life domains. Such a pattern may become evident from an examination of previous convictions, especially for violent offending, but also more broadly by establishing whether there is a pervasive tendency for the offender to attempt to impose their views of social situations and their desires on others. The extent to which other examples of self-regulatory failure are uncovered is clearly of relevance in determining the degree of focus given to self-regulation concerns in both treatment and risk management for a particular individual.

Polaschek et al. / RAPISTS’ OFFENSE PROCESSES

541

There are a number of limitations to the study. The sample was small and may not have contained a complete range of rape behaviors. Very violent and sadistic offenses were uncommon as were stereotypical date rape scenarios, stranger intruder rapes, and multiple offender scenarios. However, the model is preliminary, and accommodating change and increased diversity does not pose a problem within the grounded theory approach (Strauss & Corbin, 1990). The data collection methodology itself was limited in a variety of ways. It relied on retrospective self-reports, sometimes of events up to 5 years old. Retrospective distortion, both unintentional and intentional, is a pervasive problem in research and clinical practice in this area. The use of an interview with participant-specific prompts enhanced the level of detail compared to unprompted methods but was more vulnerable to demand characteristics and may have promoted conditions for the offender participants where actual recall was replaced by inference (Loftus, 1993). Then there are threats to veracity arising from an offender’s desire to be liked, to minimize responsibility for offending, and so on. We took no formal measures of these kinds of potential biases, and thus the model may be significantly undermined by minimization of planning and of the true antisocial nature of goals, overemphasis on external causes, and so on (see Zamble & Quinsey, 1997, for a fuller discussion of some of these points). However, we did emphasize confidentiality and the independence of the project from sentence administration, and we used independent file information to challenge offender biases when disparities were evident. Although these measures still do not give the degree of control over validity threats that would be desirable, there remain formidable obstacles to conducting research, in particular, prospective research into the covert elements of offending behavior. Nevertheless, it could be argued that the model is as much a product of what offenders want researchers and clinicians to think they think, feel, and do as it is a representation of genuine affective, cognitive, and behavioral processes. The chosen method of data analysis, grounded theory, is also open to a variety of criticisms. The early parts of construction of the model were primarily conducted by the first author, and although other investigators were involved at every subsequent stage, the methodology is likely to be vulnerable to researcher bias, which may also threaten generalizability. Thus, the next priority is to establish formally the RM’s generalizability, reliability, and validity (Henwood & Pidgeon, 1992; Rennie, Phillips, & Quartaro, 1988). It is possible to do this using a mix of qualitative and quantitative strategies. Other priorities for future research include cross-validation studies in which independent investigators collect new offense descriptions that they code and examine for fit with the existing model. These new data will extend

542

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

the size of the sample and the range of offense patterns the RM contains. We also plan to use a number of strategies for examining its conceptual validity, including documenting the pathways through the model taken by individual offenses and examining whether offender (e.g., paraphilic vs. nonparaphilic) and offense (e.g., degree of violence) characteristics vary in hypothesized ways across these pathways. We have presented here a preliminary descriptive model of the rape offense process based on verbal descriptions provided by a sample of incarcerated rapists. Although the RM needs further development, it already fills an existing gap in the research literature. It accommodates a broader range of offense pathways than the traditional RP model, distinguishes between distal and proximal goals and between sexual and nonsexual goals, and makes suggestions about the role of offenders’ interpretations of victim behavior in each phase of the offense. These results have promising implications for theory and taxonomy and program development. The construction of descriptive models helps to ensure the maintenance of sound links between the phenomena that form the basis of clinical work with rapists and the increasingly sophisticated theoretical mechanisms developed by researchers.

REFERENCES Baumeister, R. F., Smart, L., & Boden, J. M. (1996). Relation of threatened egotism to violence and aggression: The dark side of high self-esteem. Psychological Review, 103, 5-33. Bushman, B. J., & Baumeister, R. F. (1998). Threatened egotism, narcissism, self-esteem, and direct and displaced aggression: Does self-love or self-hate lead to violence? Journal of Personality and Social Psychology, 75, 219-229. Canter, D. (1994). Criminal shadows: Inside the mind of the serial killer. New York: HarperCollins. Carich, M. S. (1994, December). The use of RP/RI in sex offender treatment. III: The significance of assault cycles. Sex Offender Treatment Project Newsletter, 2-8. Darke, J. L. (1990). Sexual aggression: Achieving power through humiliation. In W. L. Marshall, D. R. Laws, & H. E. Barbaree (Eds.), Handbook of sexual assault: Issues, theories, and treatment of the offender (pp. 55-72). New York: Plenum. Freeman-Longo, R., & Pithers, W. D. (1992). A structured approach to preventing relapse: A guide for sex offenders. Orwell, VT: Safer Society. Groth, A. N., Burgess, A. W., & Holmstrom, L. L. (1977). Rape: Power, anger, and sexuality. American Journal of Psychiatry, 134, 1239-1243. Henwood, K. L., & Pidgeon, N. F. (1992). Qualitative research and psychological theorizing. British Journal of Psychology, 83, 97-111. Hudson, S. M., & Ward, T. (1996). Relapse prevention: Future directions. Sexual Abuse: A Journal of Research and Treatment, 8, 249-256. Hudson, S. M., Ward, T., & McCormack, J. C. (1999). Offense pathways in sexual offenders. Journal of Interpersonal Violence, 14, 779-798.

Polaschek et al. / RAPISTS’ OFFENSE PROCESSES

543

Knight, R. A., & Prentky, R. A. (1990). Classifying sexual offenders: The development and corroboration of taxonomic models. In W. L. Marshall, D. R. Laws, & H. E. Barbaree (Eds.), Handbook of sexual assault: Issues, theories, and treatment of the offender (pp. 23-52). New York: Plenum. Knight, R. A., Rosenberg, R., & Schneider, B. A. (1985). Classification of sexual offenders: Perspectives, methods, and validation. In A. W. Burgess (Ed.), Rape and sexual assault: A research handbook (pp. 222-293). New York: Garland. Laws, D. R. (Ed.). (1989). Relapse prevention with sex offenders. New York: Guilford. Laws, D. R. (1995). Central elements in relapse prevention procedures with sex offenders. Psychology, Crime and Law, 2, 41-53. Loftus, E. F. (1993). The reality of repressed memories. American Psychologist, 48, 518-537. Malamuth, N. M., Heavey, C. L., & Linz, D. (1996). The confluence model of sexual aggression: Combining hostile masculinity and impersonal sex. Journal of Offender Rehabilitation, 23, 13-37. Marshall, W. L., & Barbaree, H. E. (1990). An integrated theory of the etiology of sexual offending. In W. L. Marshall, D. R. Laws, & H. E. Barbaree (Eds.), Handbook of sexual assault: Issues, theories, and treatment of the offender (pp. 257-275). New York: Plenum. Marshall, W. L., Fernandez, Y. M., Hudson, S. M., & Ward, T. (Eds.). (1998). Sourcebook of treatment programs for sexual offenders. New York: Plenum. Pithers, W. D. (1990). Relapse prevention with sexual aggressors: A method for maintaining therapeutic gain and enhancing external supervision. In W. L. Marshall, D. R. Laws, & H. E. Barbaree (Eds.), Handbook of sexual assault: Issues, theories, and treatment of the offender (pp. 343-361). New York: Plenum. Pithers, W. D. (1993). Treatment of rapists: Reinterpretation of early outcome data and exploratory constructs to enhance therapeutic efficacy. In G.C.N. Hall, R. Hirschman, J. R. Graham, & M. S. Zaragoza (Eds.), Sexual aggression: Issues in etiology, assessment, and treatment (pp. 167-196). Washington, DC: Taylor & Francis. Polaschek, D.L.L., Ward, T., & Hudson, S. M. (1997). Rape and rapists: Theory and treatment. Clinical Psychology Review, 17, 117-144. Rennie, D. L., Phillips, J. R., & Quartaro, G. K. (1988). Grounded theory: A promising approach to conceptualization in psychology? Canadian Psychology, 29, 139-150. Strauss, A., & Corbin, J. (1990). Basics of qualitative research: Grounded theory procedures and techniques. Newbury Park, CA: Sage. Ullman, S. E., & Knight, R. A. (1995). Women’s resistance strategies to different rapist types. Criminal Justice and Behavior, 22, 263-283. Ward, T., & Hudson, S. M. (1998). The construction and development of theory in the sexual offending area: A metatheoretical framework. Sexual Abuse: A Journal of Research and Treatment, 10, 47-63. Ward, T., Hudson, S. M., & Keenan, T. (1998). A self-regulation model of the sexual offense process. Sexual Abuse: A Journal of Research and Treatment, 10, 141-157. Ward, T., Hudson, S. M., & Marshall, W. L. (1994). The abstinence violation effect in child molesters. Behavior Research and Therapy, 32, 431-437. Ward, T., Hudson, S. M., & Siegert, R. J. (1995). A critical comment on Pithers’ relapse prevention model. Sexual Abuse: A Journal of Research and Treatment, 7, 167-175. Ward, T., Louden, K., Hudson, S. M., & Marshall, W. L. (1995). A descriptive model of the offense chain in child molesters. Journal of Interpersonal Violence, 10, 452-472. Zamble, E., & Quinsey, V. L. (1997). The criminal recidivism process. Cambridge, UK: Cambridge University Press.

544

JOURNAL OF INTERPERSONAL VIOLENCE / June 2001

Devon Polaschek, Ph.D., Dip.Clin.Psyc., is a senior lecturer in criminal justice psychology and director of clinical training at Victoria University of Wellington, New Zealand, and research consultant to the Violence Treatment Unit at Rimutaka Prison, near Wellington. Her main research focus is on offense processes of violent offenders and rapists and their implications for assessment and rehabilitation. She is also interested in research that assists in the identification of stranger offenders in police investigations. Tony Ward, Ph.D., Dip.Clin.Psyc., is the coordinator of the forensic doctoral program at the University of Melbourne, Australia. His research interests include social competence in offenders, theories of sexual offending, cognitive distortions, and relapse prevention treatment models. Stephen M. Hudson, Ph.D., is director of clinical training at the University of Canterbury, New Zealand, and consultant psychologist to the Kia Marama Sex Offender Treatment Program at Rolleston Prison. He has been treating and carrying out research in the area of sexual aggression since 1982. He is coeditor of The Juvenile Sex Offender (with H. E. Barbaree & W. L. Marshall), Sourcebook of Treatment Programs for Sexual Offenders (with W. L. Marshall, Y. M. Fernandez, & T. Ward), and Remaking Relapse Prevention (with D. R. Laws & T. Ward). Richard J. Siegert, Ph.D., is a senior lecturer in clinical psychology at Victoria University of Wellington, New Zealand. His research interests include psychometric assessment in clinical psychology and clinical neuropsychology.