A descriptive model of dysfunctional cognitions in child molesters.

The study of cognitive factors in sexual offending has been hampered by the lack of a framework to guide model building and empirical research. There have been a number of attempts to describe the nature of offenders' beliefs and to develop ways of measuring them (e.g., Murphy, 1990; Neidigh & Krop, 1992; Salter, 1988; Segal & Stermac, 1990) but usually in the absence of any integrating theory. A consistent problem has been the tendency of studies to only focus on one or two aspects of cognition -- for example, attitudes and beliefs (Johnston & Ward, 1996; Segal & Stermac, 1990; Stermac & Segal, 1989; Ward, Hudson, & Marshall, 1994). In contrast, social cognition researchers have tended to make a useful distinction between the content (cognitive products) of cognitions and the processes that generate them. As such, there are different kinds of cognitive variables, such as cognitive structures (e.g., schemata), operations (e.g., information processing), and products (e.g., self-statements, beliefs, and attributions). Sex offenders may differ from nonoffenders on some, but not all, of these variables (Fiske & Taylor, 1991; Segal & Stermac, 1990). For example, sex offenders may have maladaptive attitudes or beliefs about the legitimacy of sex with children or forced sex with women (Abel et al., 1989; Howells, 1979; Stermac & Segal, 1989) but use adaptive (i.e., normative) information-processing strategies (e.g., confirmatory biases). Researchers in the sexual offending domain have focused most of their efforts into developing psychometrically robust measures of dysfunctional beliefs and attitudes that discriminate sex offenders from other groups (Abel, Becker, & Cunningham-Rathner, 1984; Abel et al., 1989; Hanson, Gizzarelli, & Scott, 1994). That is, the focus of this research has been on the content of relevant cognitions and its measurement.

The cognitive processes underlying the initiation, maintenance, and justification of sexual offending are just as important as cognitive content in understanding sexual offending and the development of successful treatment programs. For example, an offender may selectively attend to information, focusing on the salient behaviors of the victim that are consistent with stereotypical beliefs and attitudes. Biased attention toward these behaviors will influence subsequent judgments, and inferences as other information, particularly belief-inconsistent information, will receive correspondingly less attention. Behaviors of the victim may also be selectively interpreted in an expectancy-consistent manner, with biases particularly evident when behaviors are ambiguous. For example, if an offender believes children want sexual contact with an adult, an innocent behavior such as sitting on the offender's knee or displaying affection to him or her may be misinterpreted as indicating a willingness for sexual contact. Finally, offenders may selectively search (selective exposure) for information within their social world that is consistent with their preexisting beliefs. Gathering information in this manner excludes from processing any information that is disconfirming or inconsistent.

Finally, there is likely to be a relationship between cognitive factors and the offense chain. Models of the offense chain or relapse process (e.g., Pithers, 1990; Ward, Louden, Hudson, & Marshall, 1995) typically specify the cognitive, behavioral, motivational, and contextual factors associated with a sexual offense. These models make explicit the temporal component of offending and suggest that the functional role of cognitive distortions may change over the offense cycle (see, e.g., Ward, Louden, et al., 1995). An understanding of the cognitive processes underlying behavior at all stages of the offense chain could provide therapists with theoretical and practical guidance in the facilitation of therapeutic change. This would also provide a basis for treatment beyond ad hoc trial-and-error procedures. Research to date has typically been too static, focusing primarily on postoffense cognitions (e.g., Abel et al., 1984; Pollack & Hashmall, 1991). Although a number of cognitive processes are especially relevant for postoffense rationalizations and subsequent reoffending, information-processing mechanisms are also important before and during the offense cycle (Marlatt & Gordon, 1985; Pithers, 1990; Ward, Hudson, & Marshall, 1994). Indeed, some of these cognitive processes change markedly throughout the offending sequence as a result of increased sexual arousal and fluctuating mood states (Ward, Louden, et al., 1995).

Most of the research on cognitive factors in sexual offending has focused on the development of questionnaires (e.g., Abel et al., 1984). However, two recent studies have sought to obtain a more fine-grained description of the cognitive distortions of sexual offenders. Neidigh and Krop (1992) initiated a data-based exploration of the nature and specific content of cognitive distortions and rationalizations expressed by child molesters. Participants responded to an open-ended questionnaire on the beliefs, attitudes, and ideas that contributed to them committing an offense. More than 350 statements were generated that were then coded and reduced by card sorting to 38 categories containing equivalent content meaning. The most commonly endorsed beliefs involved victim enjoyment, absence of victim harm, or "extenuating" circumstances such as alcohol.

Second, Pollack and Hashmall (1991) described the thematic content and logical structure of excuses offered by child molesters, derived from statements extracted from case files and pretrial psychiatric assessments. The 21 distinct excuses were collapsed into six thematic categories: (a) mitigating factors, situational; (b) sex with children is not wrong; (c) incident was not sexual; (d) mitigating factors, psychological; (e) blaming the victim; and (f) denial. Victim consent, unavailability of conventional sex, drunk at the time of offense, seduction by the victim, and lack of mental control were the most common excuses offered by offenders consistent with those identified by Neidigh and Krop (1992).

Although these studies provide valuable information concerning the content of sexual offenders beliefs, they do not address the broader domain of cognitive processes, nor do they take the temporal dimension into account. We have argued elsewhere that the major role for microlevel or descriptive theories is to articulate the full domain of any phenomena under question (Ward & Hudson, 1998). The rigorous and detailed description of what actually happens, as described by offenders in this case, is required first. This data-driven description needs then to be the focus of broader, middle-level explanatory mechanisms, which in turn need to be knitted into overarching etiological frameworks. If the descriptions of phenomena that are being subjected to explanation are incomplete, then the explanations themselves are either seriously flawed or at least limited in scope.

In this research project, we set out to examine the full range of cognitive factors associated with sexual offending against children. The aim of the first study was to elicit full descriptions of offense processes and to then develop a set of descriptive categories of the cognitions child molesters use regarding their offenses and offending behavior. The aim of Study 2 was to provide a cross-validation and estimate of the reliability of classifying the descriptive categories generated in Study 1.

STUDY 1

Method

Participants

Twenty incarcerated male child molesters volunteered to participate in this research project. All participants were in the assessment phase of the Kia Marama Sex Offender Treatment Program (Hudson, Marshall, Ward, Johnston, & Jones, 1995) at Rolleston Prison, and none had undergone sexual offending treatment before. They were recruited by their therapists. Information detailing the study, the tasks involved, and participants' rights were provided, and consent was obtained in writing. Their mean age was 54.8 (SD = 11.0, range 29-65), mean length of sentence was 54.2 months (SD = 17.6, range 48-119), mean estimated number of victims was 8.5, (SD = 4.6, range 1-15), and mean length of offending was 10.6 years (SD = 9.8, range 1-40). A total of 14 men had offended against female children, 5 against male children, and I against both male and female children. Sixteen of the men were Caucasian and the remaining 4 Maori. No benefits related to incarceration or prison process were accrued by an offender for participation in this project.

Grounded Theory

Qualitative methods have traditionally received little attention in psychology because they have frequently been seen as unscientific (Henwood & Pidgeon, 1992). However, their potential contribution, particularly in the early stages of theory development, is being increasingly recognized (Rennie, Phillips, & Quartaro, 1988). We have successfully used this method previously in the sexual offending area (see, e.g., Ward, Louden, et al., 1995; Ward, McCormack, & Hudson, 1997). In this study, we used a qualitative method -- grounded theory -- to transform the interview data into categories. Grounded theory consists of a set of systematic procedures that seek to inductively derive a theory or a set of categories from qualitative data (Strauss & Corbin, 1990). Typically, concepts are inductively derived from an initial set of qualitative descriptions or scripts that, once coded into rudimentary conceptual categories, lead to the collection either of more descriptions or of quantitative data. The next step is the deduction of predictions or hypotheses concerning the ability of the provisional categories to account for new protocols. If they fail to accommodate the new data, then new categories are formulated and the process continues. Therefore, there is a progressive development of categories as the research project unfolds. The whole process of category building is dynamic and extremely sensitive to patterns detected in the data.

Procedure

During the assessment phase of the Kia Marama Sex Offender Treatment Program, each child molester undergoes various tests and interviews. The Offense History Interviews were used as a data source for …