SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Background
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Appendix

Purpose. Ward (2000) has hypothesized that sexual offenders hold offence supportive implicit theories (ITs) or schemata that function to facilitate or maintain offending behaviour. The present research aimed to determine whether rape-prone men hold the same offence supportive ITs as those that have been identified in rapists.

Method. This study adopted both an explicit measure of ITs and also an implicit measure of ITs (an interpretative bias task). In the implicit task, participants viewed ambiguous stimuli (one-sentence statements) that may be interpreted in either a rape-supportive manner, or a non-rape-supportive manner. Participant's interpretation of the stimuli was assessed via a memory recognition task. We predicted that men higher on proclivity to rape – who presumably hold strong mental representations of rape-supportive themes – would be more likely to interpret stimuli in a rape-supportive manner relative to non-rape-supportive stimuli compared to men lower on rape proclivity.

Results. Using multiple regression to determine the relative contributions of both explicit and implicit measures for predicting rape proclivity, we found that only the explicit, self-report questionnaire and one of the ITs, ‘women are sex objects’ (as measured by the interpretative bias task), was significantly related to a person's rape proclivity score.

Conclusions. This result indicates that rape-prone men may not share the same beliefs as convicted rapists, which could be a key difference between men at risk of offending, and those who have been convicted of a sexual offence.


Background

  1. Top of page
  2. Abstract
  3. Background
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Appendix

A great number of researchers and clinicians have noted that sexual offenders frequently make comments or statements that appear to support or justify either sexual offending in general, or their own sexual offending behaviour. These statements – often termed ‘cognitive distortions’ (CDs) in the literature – are hypothesized to play some role in the offence process (Johnston & Ward, 1996; Ward, Hudson, Johnston, & Marshall, 1997). However, despite the significant research interest in this area, there is much conjecture about the exact definition of CDs. The pioneers of the investigation of CDs in child molesters, Abel, Becker, and Cunningham-Rathner (1984), defined CDs as belief systems that support sexual offences and also act as justifications, perceptions, and judgments that may be used by the offender to rationalize their offending behaviour. Other researchers, however, propose that CDs might not be reflecting a belief system at all, and instead may only be statements that are made after the offence, in an attempt to either justify, excuse, or rationalize behaviour to others (see Maruna & Mann, 2006), or themselves, in an attempt at self-deception (Gannon & Polaschek, 2006). Until relatively, recently researchers have attempted to assess CDs using self-report measures such as The Abel and Becker Cognitions Scale, ABCS; Abel et al., 1989). However, this methodology may not differentiate between offence supportive beliefs held by offenders, or statements that are made in an attempt to rationalize behaviour, making it difficult for researchers to draw firm conclusions about the role CDs play in the offence process. Ward (2000), arguing that cognition plays an aetiological role in sex offending, proposed a new theory using information processing theory as a basis for explanation.

Information processing theory postulates that differences in the way information is stored and organized in individuals’ long-term memory (as schemata) biases attention, encoding, and retrieval of new information, therefore affecting subsequent behaviour (Fiske & Taylor, 1991). The way that information is stored and organized in memory depends on early life experiences, and will vary between individuals. Ward (2000) suggested that schemata should be regarded as causal theories that interact with information from personal experiences to form coherent cognitive structures that are used to both explain and predict our own behaviour, and that of others. Ward termed these theories ‘implicit theories’ (ITs), and with colleagues, began to examine the possibility that offenders hold specific offence-supportive schemata that may facilitate and reinforce offending behaviour (Polaschek & Ward, 2002; Ward & Keenan, 1999).

Polaschek and Ward (2002) proposed five specific ITs for rapists; ‘Women are unknowable/dangerous’ (beliefs that men are unable understand or predict a woman's behaviour), ‘Women are sex objects’ (beliefs that the only purpose of women is to please a man sexually), ‘Male sex drive is uncontrollable’ (beliefs that men have no control over their sexual arousal, and need to be satisfied sexually when this happens), ‘Entitlement’ (beliefs that all men are entitled to sex), and ‘Dangerous world’ (beliefs that the world is a hostile place). According to Ward, these ITs may create processing biases when an individual encounters a scenario that is inconsistent with the stored knowledge of the schemata – or is outside of their own previous experience. In these situations, it is suggested that the information is encoded, processed, and interpreted in accordance with the schemata, and not the actual events. Polaschek and Ward (2002) propose that individuals holding these beliefs may be prone to misattributing sexual intent to non-sexual behaviour. For example, a woman may be seen as dressing in a particular way specifically to attract sexual invitations, creating a dangerous situation in terms of potential sexual offending.

Rape proclivity

In studying factors that contribute to the aetiology of rape, researchers have recognized the need to study not only incarcerated rapists, but also community males identified as having some propensity to sexually aggress. Several self-report questionnaires have been designed to measure the extent to which men demonstrate a proclivity to rape – that is, a likelihood to rape (e.g., The Likelihood to Rape index, Malamuth, 1981; The Attraction to Aggression Scale, Malamuth, 1989; The Sexual Experiences Survey, Koss, Gidycz, & Wisniewski, 1987; and The Rape Proclivity Measure, Bohner et al., 1998). A large amount of research has been conducted with rape-prone men (e.g., Bohner et al., 1998; Bohner, Siebler, & Schmelcher, 2006; Malamuth, 1981; Malamuth & Check, 1980; Malamuth, Haber, & Feshbach, 1980) with studies suggesting that men who score high on rape proclivity tend to endorse more myths or beliefs that are accepting of rape (see Bohner et al., 1998; Malamuth & Check, 1985) and these men share similar characteristics to convicted rapists, such as high levels of hostility (Malamuth, 1986), and feelings of anger towards women (Lisak & Roth, 1988). This body of research offers support for the theory that data collected on rape-prone men can be applied or generalized to convicted rapists. If this is the case, then investigating ITs in rape-prone men could be useful for furthering our understanding of CDs or rape-supportive beliefs.

Ward and colleagues’ theory of offence-supportive schemata provides an excellent framework for investigation in this area. If, as proposed by Ward and colleagues, offenders hold specific offence-supportive schemata, that they use to interpret the world around them, then it should be possible – through the use of cognitive methods – to implicitly examine the way these individuals’ attend, encode, and interpret social information they receive. A large proportion of studies that have attempted to examine rapists’ beliefs was designed before the proposal of IT theory, and as such use explicit measures. For example, Bumby (1996) developed a generic scale of rape-supportive beliefs (Rape Scale) and found that rapists’ responses on this scale did differentiate them from offenders who had not sexually offended. There is also a small body of research using samples of rapists or rape-prone men that has been successful in identifying ITs (e.g., Polaschek & Gannon, 2004; Polaschek & Ward, 2002). For example, Polaschek and Gannon (2004) analysed the offence accounts of 37 convicted rapists and found strong evidence for all five ITs, the most prevalent being ‘women are sexual objects’, ‘entitlement’, and ‘women are unknowable/dangerous’ occurring in 70%, 68%, and 65% of interview transcripts, respectively. ‘Male sex drive is uncontrollable’ and ‘dangerous world’ also occurred in a minority of cases. This study is an important first step in the investigation of ITs in rapists, but the data collected relies on self-report information from the offenders, which is open to social desirability bias. Furthermore, we cannot draw any conclusions about the role these ITs play in the offence process, as we cannot distinguish the root cause or function of these statements.

As mentioned previously, employing information processing measures – implicit measures – may be useful for investigating CDs for two reasons. First, this methodology does not directly ask the participant for information, therefore reducing social desirability biases. Second, such a methodology can measure attitudes or processes that the participants themselves are not consciously aware of (see Fazio & Olson, 2003).

Although studies have begun to examine rapists’ ITs using self-report measures (e.g., Polaschek & Gannon, 2004), there has been little attempt to investigate the presence of these ITs using implicit, cognitive methodology in either rapists or rape-prone men – despite the fact these methods have been utilized successfully to examine ITs in child molesters (e.g., Gray, Brown, MacCulloch, Smith, & Snowden, 2005; Kamphuis, De Ruiter, Janssen, & Spiering, 2005; Keown, Gannon, & Ward, 2008; Mihailides, Devilly, & Ward, 2004). The examination of rape-prone mens’ cognition is useful for furthering our understanding about men who are at risk of offending as well as those who have committed offences, as we may be able to generalize these findings to convicted rapists – a sample that is often difficult to obtain. Furthermore, studying a group of men who, theoretically, are at risk of offending – and who hold the same belief systems as convicted rapists – may help us to further understand the potentially protective factors that are inhibiting these men from committing sexual offences.

Previous attempts have been made to examine the five ITs thought to be held by rapists in a sample of rape-prone men. Blake and Gannon (2010) used a lexical decision task (LDT) to investigate whether men who score higher on Bohner et al.'s (1998) Rape Proclivity Measure were quicker to respond to rape supportive, IT consistent words, compared to men obtaining lower scores on the Rape Proclivity Measure. In this study, it was predicted that response times (RTs) from the LDT would be more representative of the presence of ITs because an implicit, cognitive measure of ITs should, theoretically, be more appropriate for examining ITs than explicit methods. To test this hypothesis, the authors also implemented an explicit measure (The Rape Scale: Bumby, 1996) in order to calculate the relative predictive utility of each method using regression analysis. Although the regression model was significant, only 13% of variability in rape proclivity score could be predicted from the independent variables collectively (Rape Scale score, and the five ITs as measured by the LDT). Furthermore, the Rape Scale was the only significant predictor variable. For three of the five ITs (women are unknowable, women are sex objects, and male sex drive is uncontrollable), rape-prone men tended to respond faster to the IT consistent stimuli than non-rape-prone men, but this pattern was not significant. For the remaining two ITs however, (entitlement and dangerous world), this pattern was reversed (but also not significant), with rape-prone men responding faster to the non-IT consistent stimuli. This result makes it difficult to draw any firm conclusions about the ITs or rape-supportive beliefs in rape-prone men. Either rape-prone men do not hold the same ITs as rapists, or the LDT methodology employed did not tap into the ITs as expected. Such a null finding could also be accounted for by low power, due the relatively small sample size employed.

The present study

The present study utilizes the interpretative task paradigm (see Gannon & Rose, 2009) to examine three of the ITs thought to be held by rapists in rape-prone men. The interpretative task is an adaptation of a memory recognition task that has been used previously in clinical and forensic populations (Copello & Tata, 1990; Eysenck, Mogg, May, Richards, & Matthews, 1991; Gannon & Rose, 2009). The main assumption of this paradigm is that ambiguous stimuli will be interpreted and therefore subsequently recognized in a manner consistent with schemata. Thus, in this particular task, participants view stimuli that may either be interpreted in a rape-supportive or non-rape-supportive manner. Then, participant's interpretations of the stimuli are examined through participants’ recall of the stimuli. In line with information processing theory, and Ward's IT theory, we predict that men scoring high on rape proclivity will interpret the original stimuli in a rape-supportive manner, due to the ITs that they hold.

The participants also complete an explicit self-report measure of rape-supportive cognition (Bumby's Rape Scale).

Based upon the research evidence to date, we make two main predictions. First, we predict that men who score higher on the Rape Proclivity Measure will endorse more rape-supportive beliefs on the Rape Scale relative to those who score lower on rape proclivity. Second, we predict that men who score higher on the Rape Proclivity Measure will show a pattern of greater recognition for rape-supportive stimuli than non-rape-supportive target stimuli relative to men obtaining lower scores on the Rape Proclivity Measure, and would be faster to make these rape-supportive recognitions. Finally, to compare explicit and implicit measures of rape-supportive cognitions, we will use a regression model consisting of each participants’ performance on the interpretative bias task and Rape Scale score to assess the relative contribution of implicit and explicit measures for predicting rape proclivity score.

Method

  1. Top of page
  2. Abstract
  3. Background
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Appendix

Participants

Participants were 78 males aged between 18 and 37 years (mean age 21.09 years, SD= 3.40) who volunteered to take part in a study titled ‘Memory recall study’. Eight participants’ data were removed from the study as these males reported their sexual orientation as either homosexual or bisexual.1 Participants were recruited through advertisements on research participant websites and on the University's student job page. Participants could choose to receive either £5 for taking part, or five credits towards fulfilment of an undergraduate psychology course. Participants were primarily University students (93.5%). All participants had spent a minimum of 14 years in formal education.

Materials

Rape Proclivity Scale

The Rape Proclivity Measure (Bohner et al. 1998) requires participants to read five realistic date rape scenarios, and asks participants to imagine themselves in the position of the male protagonist. Participants then answer the following three questions with respect to each scenario. ‘In this situation, how aroused would you be?’ (1; not at all sexually aroused, to 5; very strongly sexually aroused); ‘In this situation, would you have done the same?’ (1; would definitely not done the same, to 5; would definitely have done the same), and ‘In this situation, how much would you enjoy getting your way?’ (1; would not enjoy it at all, to 5; would greatly enjoy it). Scores on this scale range from 15 to 75, although Bohner and colleagues sum questions two and three across all five scenarios to create the measure of rape proclivity, with a range of 10 to 50. The Cronbach's alpha of the combination of these two questions is α > 80. Bohner and colleagues have demonstrated that this measure is unaffected by social desirability r (111) = .05, p > .61 and in addition found that the measure correlates positively with men's self-reports of past sexual aggression, r (112) = .38, p < .001.

Rape Scale

The Rape Scale (Bumby, 1996) – an explicit measure of rape-supportive beliefs – consists of 33 statements, each followed by a 4-point Likert scale on which to rate agreement that excludes a ‘neutral’ response option. Examples of questions are ‘Men who commit rape are probably responding to a lot of stress in their lives, and raping helps to reduce that stress’ and ‘Woman generally want sex no matter how they can get it’. The Rape Scale has excellent psychometric properties (internal consistency α= .96, test–retest reliability r= .84; Bumby, 1996). A 5-point Likert scale was used for the purpose of this study to ensure respondents had a neutral response option to rate their responses (as requested by the Ethics Board).

The interpretative bias task

The computerized interpretative bias task presented participants with 18 sentences and corresponding sentence derivatives in the recognition phase that were designed to represent three of the five ITs identified by Ward and Polaschek (women are dangerous, women as sex objects, and male sex drive is uncontrollable). These three ITs were chosen for investigation following previous research (Blake & Gannon, 2010), in which the findings (though non-significant) suggested that rape-prone men might be more likely to hold these ITs. Participants were also presented with 10 control sentences, taken from Gannon and Rose (2009), that were designed to assess generally negative social interpretations, and a further 10 ambiguous filler sentences (also taken from Gannon & Rose, 2009) designed to disguise the true aims of the task.

Of the 18 sentences designed to assess interpretation of rape-supportive schemata, six represented the IT women are unknowable, six represented the IT women are sex objects, and six represented the IT male sex drive is uncontrollable. The full list of stimuli can be found in the Appendix.

During the encoding phase, the presented sentences remained on the screen until the participant pressed the space bar to move on to the next sentence. In the recognition phase, participants were given instructions to read the sentences presented to them and decide whether they recognized the meaning of the sentence from those they had been shown previously.

Stimuli

An ANOVA confirmed that sentence length (number of characters per sentence) did not differ significantly across sentence type (original ambiguous sentence, rape-supportive sentence derivative, and non-rape-supportive sentence derivative), F(2, 81) = .017, p= .98. In addition, ANOVAs were performed individually for each IT and the control sentences. The ANOVAs confirmed that sentence length did not differ significantly across sentence type for all three ITs. Table 1 shows mean sentence length for all sentence types used in the interpretative bias task.

Table 1. Mean number of characters per sentence type
Sentence typeMean number of characters per original sentencesMean number of characters per rape-supportive sentencesMean number of characters per non-rape-supportive sentences
Women are unknowable33.673234
Women as sex objects32.1733.1734.5 
Male sex drive is uncontrollable5050.8347.83
Control sentences49.5 50.9 50.6 

Two versions of the interpretative bias task were implemented – in line with previous research using this procedure (e.g., Copello & Tata, 1990; Gannon & Rose, 2009) – so that participants only saw one derivative of each original sentence. For example, of the six original sentences representing Women are unknowable, each participant would see three rape-supportive interpretations and three non-rape-supportive interpretations in the recognition phase.

Apparatus

The interpretative task was created using the computer software E-Prime. Written instructions and all sentences were presented in black, Times New Roman text on a white background. Participants made their responses using the computer keyboard and RTs were recorded in milliseconds.

The E-Prime programme controlled the random presentation of original ambiguous sentences in the encoding phase, and the sentence remained on the screen until the participant pressed a key to move on to the next sentence. In the recognition phase, the E-prime programme controlled the random presentation of an even number of rape-supportive, and non-rape-supportive sentence derivatives. Participants responded to the stimuli by pressing one key if they recognized the sentence meaning, or another key if they did not recognize the sentence meaning. Participants were told to respond as quickly as possible. The type of response made (recognized/not recognized) and RT (in milliseconds) was recorded by the programme.

Procedure

Participants were invited to a psychology lab where they were given an information sheet to read, which explained what they would be asked to do, and informed them that their responses would be anonymous and that they had the right to withdraw from the study at any time without penalty. Participants who were satisfied with this information signed a consent form to agree to take part and to demonstrate that they understood the procedure. Participants were tested individually. Participants were given verbal instructions on how the session was going to run (i.e., that they would complete questionnaires before the computer task or vice versa, due to counterbalance procedure). When participants had successfully completed all stages of the experiment they were debriefed. Ethics approval was provided by the authors’ University Ethics Board.

Results

  1. Top of page
  2. Abstract
  3. Background
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Appendix

Table 2 shows descriptive statistics and correlations for the dependent variable (rape proclivity score) and all independent variables.

Table 2. Descriptive statistics and correlations for dependent variables and all independent variables
VariablesRape proclivityWomen are unknowableWomen are sex objectsMale sex drive is uncontrollableRape Scale
  1. *p < .05; **p < .01.

Women are unknowable−.11    
Women are sex objects.29*−.01   
Male sex drive is uncontrollable−.16.00.01  
Rape Scale.70**−.22.21.00 
Means17.43−.39−.37−.3368.00
Standard deviations6.221.031.081.1114.86

Rape proclivity was calculated by adding scores for questions 2 and 3 together across all five scenarios, as in Bohner et al. (1998). The final score therefore indicated the extent to which participants could imagine themselves acting in the same way as the male protagonist in the date rape scenarios, and how much they would enjoy getting their own way in such a situation. Scores could range from 10 to 50. In this sample, the mean score was 17.43 (SD= 6.22) with a range of 10–34, indicating slight floor effects, but still representing a sizable range of scores. Cronbach's alpha of the combination of these two questions in this sample is α > 70. To test the first hypothesis, that participants who score higher on the Rape Proclivity Measure will score higher on rape-supportive beliefs as measured by the Rape Scale, relative to lower scorers on the Rape Proclivity Measure, we conducted a correlational analysis. As predicted, the two variables were significantly correlated, r= .70, p < .001, demonstrating a large effect size (49% shared variance).

Recognition analysis

The difference between number of rape-supportive sentences and number of non-rape-supportive sentences was calculated to create one score that reflected recognition of rape-supportive sentences over and above recognition of non-rape-supportive sentences. This score was calculated for each IT, meaning each participant had three such scores (plus one score for control sentences). Positive scores represent greater recognition of rape-supportive stimuli, and negative scores represent greater recognition of non-rape-supportive stimuli.

The mean recognition scores of –.39 (SD= 1.03) for women are unknowable, –.37 (SD= 1.08) for women are sex objects, and –.33 (SD= 1.11) for male sex drive is uncontrollable sentences indicate a general response bias to non-rape-supportive stimuli.

To test the second hypothesis, that participants who score higher on the Rape Proclivity Measure will recognize more rape-supportive sentences than non-rape-supportive sentences, correlational analyses were performed. As shown in Table 2, only the women are sex objects IT was positively correlated with rape proclivity (r= .29, p < .05). Neither women are unknowable or male sex drive is uncontrollable was significantly correlated with rape proclivity (r=−.11, p= .37 and r=−.16, p= .18, respectively. Furthermore, the direction of the relationship between these variables was contrary to predictions – the pattern of responding suggests that higher scorers on rape proclivity were related to more non-rape-supportive recognitions than rape-supportive recognitions.

Finally, to determine the relative contributions of responses on the interpretative bias task (the implicit measure) and Rape Scale (the explicit measure) in predicting rape proclivity scores, a standard multiple regression was performed. The regression model consisted of four independent variables; recognition score for each IT plus the score of rape-supportive cognitions as measured by the Rape Scale. In this method, all independent variables are entered into the regression equation at once. The dependent variable consisted of rape proclivity score. Analysis and evaluations of assumptions were performed using SPSS 17. No assumptions were violated, and none of the variables were transformed.

Table 3 displays the standardized regression coefficients (β), and adjusted R2.

Table 3. Variables predicting rape proclivity score
Variableβ t p
  1. Adjusted R2= .52, df = 4,69, F= 19.81, p < .01.

Women are unknowable.04.44.66
Women as sex objects.192.30.03
Male sex drive is uncontrollable−.16−1.93.06
Rape Scale.687.83.00

R for regression was significantly different from zero, F(4, 69) = 19.81, p < .001, with R2 at .74. The adjusted R2 value of .52 indicates that a substantial amount of the variability (52%) in rape proclivity score is predicted by the three ITs (as measured by the interpretative bias task) and scores on the Rape Scale. However, only two regression coefficients differed significantly from zero; Rape Scale (t= 7.83, < .001, β= .28) and recognition for women are sex objects sentences (t= 2.30, p < .05, β= 1.12).

Altogether, 52% of the variability in rape proclivity was predicted by knowing the score on the Rape Scale and recognition of rape-supportive sentences representing the three ITs. The direction of the relationship indicates that men who score higher on the Rape Proclivity Measure are more likely to endorse rape-supportive statements as measured by the Rape Scale and interpret ambiguous stimuli as consistent with the IT women are sex objects IT. Scores on the Rape Scale were the best predictor of rape proclivity, with these scores accounting for 44% of the variance in rape proclivity compared to the 3.7% accounted for by women are sex objects IT. For the two regression coefficients that significantly differed from zero, 95% confidence limits were calculated. The confidence limits for Rape Scale were 0.203–0.342, and those for women are sex objects were 0.292–2.139.

RT analysis

RTs were windsorised so that outlier RTs falling ±two standard deviations from the grand mean were modified to the next most extreme value (Ratcliff, 1993).

The difference between RTs to rape-supportive sentence recognitions and RTs to non-rape-supportive sentence recognitions was not correlated with rape proclivity score for any of the three ITs. Thus, contrary to our predictions, rape-prone men did not demonstrate accelerated RTs relative to non-rape-prone men for their recognition of rape-supportive sentences.

Discussion

  1. Top of page
  2. Abstract
  3. Background
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Appendix

It was predicted that men scoring higher on rape proclivity would also score higher on the Rape Scale relative to men who obtained lower scores on the Rape Proclivity Measure. The significant correlation between these variables confirms that this is the case, indicating that men reporting a higher proclivity towards committing rape endorse more rape-supportive statements as measured by the Rape Scale. This result is in line with previous research indicating that rape-prone men tend to endorse rape-supportive statements on questionnaires of rape-supportive beliefs compared to non-rape-prone men (Blake & Gannon, 2010; Bohner et al., 1998; Malamuth & Check, 1985).

Further, it was predicted that men who obtained higher scores on the Rape Proclivity Measure would show a pattern of greater recognition for rape-supportive sentences than non-rape-supportive sentences relative to those obtaining lower scores on the Rape Proclivity Measure. However, a significant positive correlation between IT recognition patterns and Rape Proclivity was found for only one of the ITs measured –women are sex objects (p < .05). The pattern of responding to the other two ITs was contrary to predictions, with high scorers on Rape Proclivity more likely to make positive recognitions of non-rape-supportive sentences, although these correlations were not significant.

The regression model was designed to examine whether implicit recognition scores to each individual IT in the interpretative bias task together with the explicit questionnaire measure of rape-supportive beliefs could be used to predict an individual's rape proclivity score. Although the model did significantly predict 52% of the variance in rape proclivity scores, the fact that the Rape Scale and recognition of the women are sex objects consistent sentences were the only two independent variables to significantly contribute to the model is slightly disappointing. This means that although overall the model does help to predict a large amount of the variance in Rape Proclivity scores, two of the three individual ITs (as measured via the interpretative bias task) did not significantly contribute to this variance.

Although other researchers have successfully identified ITs in child molesters using implicit methods (e.g., Kamphuis et al., 2005; Keown et al., 2008; Mihailides et al., 2004), these methods have yet to be used with rapists (to our knowledge), and an investigation of the ITs in rape-prone men (Blake & Gannon, 2010) was unsuccessful in identifying ITs using cognitive methods. There are several possible explanations for our findings; rape-prone men may not hold the same ITs as rapists, the interpretative bias task may be failing to tap in to the ITs, or our sample may have not been large enough to catch small effects (e.g., power calculations indicate that a sample of 934 would be required to detect a small effect). Furthermore, it is possible that different individuals hold different combinations of these three ITs. First, individuals may not hold all three schema studied, and instead may hold only one, or two of the proposed schema. Thus, when results are averaged across individuals, this variance may be lost.

It is interesting that the self-report measure of rape-supportive cognitions was one of only two significant predicator variables, accounting for the highest proportion of variance, when we anticipated the implicit measure of the interpretative to be a more robust measure of men's beliefs. As in our previous work we suggest that this may be because rape-prone men hold more specific rape-supportive cognitions – as measured by the Rape Scale – compared to the more general offence related beliefs described by ITs. For example, sentences used in the interpretative bias task to describe the IT male sex drive is uncontrollable included ‘If a man fancies a woman, he won't be able to control his urges’, which refers, very broadly, to all men. In contrast, the Rape Scale contains such statements as; ‘Most of the men who rape have stronger sexual urges than other men’, which clearly describes how a rapist differs in his desires from other men. As our sample of rape-prone men are unlikely to consider themselves as rapists, it may be they hold views about rapists that they are willing to endorse visa the Rape Scale, but do not hold similar views about men in general. It appears therefore that the Rape Scale is measuring different rape-supportive beliefs from those beliefs represented by Ward and colleagues five ITs.

This difference in the type of beliefs endorsed by rape-prone men and convicted rapists may be linked to protective factors that inhibit sexually aggressive behaviour. Rape-prone men are considered to demonstrate a proclivity towards sexual aggression, but, to our knowledge have not been convicted of such offences. As it has been postulated that offence-supportive schema can facilitate sexual offending, then a lack of all five ITs in rape-prone men may be inhibiting sexual offending. The fact that we found evidence for women as sex objects, but not male sex drive is uncontrollable, nor women are unknowable, may indicate that an ability to control sex drive and an absence of negative beliefs about women are protective factors against sexual offending. Clearly, differences between these populations such as these need to be examined more thoroughly in order for us to draw stronger conclusions however.

A further limitation concerns the contextual environment of the study. Ward's (2009) extended mind theory of CDs in sex offenders posits that sex offenders may not demonstrate distorted thinking, or offence-supportive beliefs, in every situation, and instead these beliefs are likely to be context dependent. Extended theory of mind proposes that individuals’ utilize both internal and external cognitive resources when engaged in cognitive processing (Menary, 2007), and therefore a number of these resources will be situational, or contextually dependent. Therefore, in the present study, when engaged in the cognitive processing required by the recognition task, or responding to the Rape Proclivity Measure scenarios, participants may not demonstrate the same cognitive processes that may exist during an actual interaction with a female, or when engaged in sexual activity.

As mentioned earlier, another possibility for the non-significant results of this study may be due, in part, to the methodology employed. The effectiveness of interpretative bias task relies on strong stimuli that accurately represent the belief systems of interest. It is possible that our stimuli were not representative enough of the rape-supportive beliefs. This does not however explain the predictive utility of the women are sex objects IT. The fact that this IT has significant predictive power for rape proclivity demonstrates that the interpretative bias task and stimuli may have some validity. It is possible however that the women are sex objects stimuli were more accurately representative of this IT, relative to the other two stimuli sets, resulting in successful identification of this IT only.

Conclusion

Our findings suggest that an explicit measure of rape-supportive beliefs and the implicit measure of the IT women are sex objects could significantly predict rape proclivity score. However, because the implicit measure of the two other ITs –women are unknowable and male sex drive is uncontrollable– did not contribute significantly towards the prediction of rape proclivity, it is difficult to draw any firm conclusions about the nature of ITs in rape-prone men. It could be that rape-prone men do only hold one of the ITs proposed to be held by rapists, or as discussed it may be possible that the interpretative bias task is not accurately identifying the presence of all ITs. The explicit measure is substantially correlated with rape proclivity, indicating that men scoring higher on rape proclivity certainly do endorse rape-supportive beliefs on some level, however due to the self-report methodology, we cannot make any inferences about the rape-supportive cognition that may underlie these beliefs. It would be very beneficial if we could extend this research to include a sample of incarcerated rapists, to see whether the interpretative bias task is successful in measuring ITs in men known to have committed offences. This would help to both test the validity of the task, and also learn more about the differences between rape-prone men and rapists. To explore these differences further it would also be useful to examine a wider community control sample, as our participants were primarily university students, which reduces the generalizability of our findings. For example, Blake and Gannon (2010) have suggested that this type of sample is not very diverse, as university students may have not been exposed to overtly sexual or hostile female figures leading to development of these ITs. Indeed, campus life lends itself to positive interpersonal interactions between sexes, which could also account for a lack of IT consistent schema. Finally, the lack of a consensus between the self-report measure of rape-supportive beliefs and ITs could be further examined through research. As discussed previously, the self-report measure, the Rape Scale, appears to be measuring very different beliefs than those described by Ward and colleagues ITs, and yet, despite potential social desirability problems still remains a valid and reliable measure, outperforming the implicit measure in the present research. This unexpected result throws up many questions regarding the types of rape-supportive beliefs held by rape-prone men, and also the utility and validity of implicit measures.

Footnotes
  • 1

    These participants’ data were removed because the Rape Proclivity Measure only describes instances of rape between heterosexual dyads.

References

  1. Top of page
  2. Abstract
  3. Background
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Appendix

Appendix

  1. Top of page
  2. Abstract
  3. Background
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Appendix
Table Appendix:. Interpretative Bias Task Stimuli
Implicit theoryAmbiguous sentenceIT consistent sentenceNon-IT consistent sentence
Women areJohn practiced mind reading with SarahJohn tried to read Sarah's mindJohn rehearsed magic tricks with Sarah
 unknowableSometimes having a relationship with a woman can be intriguingSometimes having a relationship with a woman can be dangerousSometimes having a relationship with a woman can be fun
 Julie decided that she and Mike should take a breakJulie decided that she and Mike should break upJulie decided that she and Mike should take a tea break
 Tom was driven from his home by AnneTom was pushed out of his home by AnneTom was driven in Anne's car from his home
 Mike thought Jane was funnyMike thought Jane was weird.Mike thought Jane was amusing
 Jen came across as cunningJen came across as sneakyJen came across as clever
Women are sexSue was always prepared for anythingSue was always ready to have sexSue was always ready for problems
 objectsMary always liked fooling around with TimMary liked to have sex with TimMary liked to have fun with Tim
 Luke noticed that Jenny was looking hot todayLuke noticed that Jenny was looking sexy todayLuke noticed that Jenny was feeling the hot weather
 When Sally went to the bar she was looking for a good timeWhen Sally went to the bar she was planning on having sexWhen Sally went to the bar she was planning on enjoying the evening
 Ruth was always up for anythingRuth was always willing to do anything sexuallyRuth was always willing to try new things
 Some women are dirtySome women are promiscuousSome women are unhygienic
Male sex drive is uncontrollableMike couldn't help but be excited at the thought of meeting LauraMike couldn't help being turned on at the thought of meeting LauraMike couldn't help but look forward to meeting up with Laura
 David was embarrassed at his response to the underwear modelDavid was embarrassed that he got aroused when he saw the underwear modelDavid was embarrassed about what he said to the underwear model
 John found it difficult to control his feelings towards Kate.John found it difficult to control his sexual urges towards KateJohn found it difficult to control his emotions about Kate
 If a man fancies a woman he won't be able to control his behaviourIf a man fancies a woman he won't be able to control his urgesIf a man fancies a woman he won't be able to control his emotions
 If a man is attracted to a woman, he wants to get her attentionIf a man is attracted to a woman, he wants to get her in to bedIf a man is attracted to a woman, he wants to get to know her
 A man's sexual behaviour is ruled by his thoughtsA man's sexual behaviour is ruled by his desireA man's sexual behaviour is ruled by his head