text in 
English (pdf)
Article in xml format
Article references
Automatic translation
Permalink
A sex offender (SO) against children can be defined as an individual who practices child sexual violence, such as sexual acts or games in which a child is forced to participate, and who has more advanced psychosexual development than the child. In this way, the SO uses a power relationship to satisfy his own desires to the detriment of the victimized child’s well-being (Spaziani & Maia, 2015).
Considering the sociodemographic data of Brazil, SO are usually male1, aged between 30 and 40 years, mostly convicted of only sexual crimes, and are commonly close to his victims, such as fathers, stepfathers, uncles, neighbors, etc.). The SO prefers child and adolescent victims, and, as a consequence of proximity, he chooses the victim’s house or his own as the place for the sexual abuse (Soares et al., 2016; Teixeira et al., 2020).
Studies in the area of psychology, more specifically in the assessment of personality, carried out with SO are not numerous. Considering the projective tests that assess the personality, the Rorschach’s has been one of the most used, accepted and requested in the practice of forensic psychological assessment. Its use is justified because it has the potential to reveal personality traits that people do not fully recognize in themselves or hesitate to mention when asked about them directly. Therefore, the instrument is less susceptible to conscious and intentional manipulation or concealment by the examinee (Gacono & Evans, 2011; Nørbech et al., 2016).
Zilki et al. (2020) carried out a systematic review, covering the previous ten years of studies that investigated the personality of SO using the Rorschach test, searching on the Web of Science, PsycNET, Periódicos Eletrônicos em Psicologia (PePSIC), and Scientific Electronic Library Online (SciELO) databases. The results showed no single pattern with regard to the cognitive and affective aspects of self-perception and impulse control of SO. These divergences may have occurred due to the fact that 44% of the studies were of the case study type.
However, the authors observed that most of the investigations listed showed cognitive distortions but not necessarily the same type of distortion (Xu%↑, X-%↑, L↑, WSum6↑, and PTI+), with the most simplistic and superficial information processes, less conventional perceptions, and arbitrary and distorted thoughts predominating. Considering the impairments in self-perception and in interpersonal relationships, the most frequently cited were immature perception of oneself and the other person, distant interpersonal relationships, and a marked tendency to overestimate one’s own value (Fr+rF↑, MOR↑, FD + SumV = 0, Htotal↓, H = 0, PER↑, SumT = 0). Regarding the affective aspects, affective withdrawal, emotional lability, absence of signs of remorse, and guilt were highlighted, as well as a greater predisposition to anger and resentment (Afr, CF↑, V = 0, S↑).
Finally, regarding impulse control, the most frequently raised aspects in the studies found were few efficient resources to deal with stressful situations (EA↓), empathic inabilities and less willingness to think before acting (M↓), propensity to self-gratifying behaviors when faced with instinctive pressures (FM = 0), and greater aggression (AGC). It should be highlighted that all these impairments corroborate the predisposition for the SO not only to commit the sexual crime but also to repeat this behavior (Zilki et al., 2020).
In the legal context, dangerousness is defined as the potential or probability (not mere possibility) of a person practicing conduct prohibited by society, considering their previous acts or the circumstances in which they committed a crime. It would be the state of someone who can expose someone else to danger or cause them harm, which is determined by any mental illness that takes away or restricts their judgment or by their own actions or particular inclinations toward malevolence or the practice of it is the evidence of (or results). Criminal dangerousness is the risk that the person will be violent again, it is evidence of (or results from the practice of) the crime (Slaibi & Gomes, 2014), often related to the severity of the abuse and the crime history of the offender.
Conceptually, there is a distinction between the probability that an individual will commit another crime in the future (risk of recidivism) and the level of dangerousness or adverse consequences for the victim of a crime (risk of dangerousness). Therefore, a person may be judged to have a high risk of recidivism, but the intended act may be stealing. In turn, a person may have a low risk of criminal recidivism, but, if the crime is committed, significant consequences are expected for the victim. A good risk measurement assesses both the risk of recidivism and the risk of dangerousness that the individual presents to society (Sheldon & Howells, 2017).
In the case of SO, Rowlands et al. (2017) and Riquelme et al. (2004) highlighted that the risk of recidivism of sexual crime is determined by several factors, among them: the age of the SO, his life history, history of arrest for sexual and non-sexual crimes, his response to psychological treatment programs, substance misuse, proximity to the victim, and the presence of psychopathological disorders, such as antisocial characteristics and psychopathy. There are also patterns of violence associated with factors, such as: offensive sexual behavior, lack of control of affects and mood (such as anger and the lack of cooperation with supervision requirements), deviant sexual interests, sexual self-regulation problems, and distorted attitudes or beliefs that favor the sexual offense. In the case of SO against children, certain characteristics of the victim are included, such as age and sex (Stinson & Becker, 2016).
In the United States, the perception that people who were arrested by sex crimes has a bigger dangerousness degree caused a lot of federal and state strict laws, including the begin of the register of the sex abusers, community notifications, address requirement, and DNA collect (Gibbs & Ten Bensel, 2021).
As a consequence, several studies with different focus were carried out to understand the level of dangerousness of the SO. One of them was developed by Langton et al. (2006), which analyzed the relationship of recidivism risks in SO who underwent cognitive-behavioral psychological treatment programs. The results showed that 36.5% of the study sample recommitted a sexual crime within a mean period of five and a half years after serving the sentence of a previous sexual crime.
The study carried out by Riquelme et al. (2004), with SO and criminals sentenced for other types of non-sexual crimes evaluated with the Psychopathy Checklist-Revised (PCL-R) to identify psychopathy, indicated that SO presented significantly more affective and interpersonal characteristics of psychopathy than other criminals (p = 0.007), showing a higher probability of sexual crime recidivism. The study of these authors corroborates others that argue that deviant sexual behavior, associated with psychopathy, has been shown to be a strong ally for criminal recidivism and dangerousness.
It is highlighted that the psychopathy has been considered psychopathological disorder, hard to remit and treat and with high levels of criminal recidivism (Balsis et al., 2017; Gacono et al., 2011). Regarding comparisons among SO with and without psychopathy, those whose psychopathy (PCL-R ≥ 30) present less stress than those without psychopathy (PCL-R < 30) in the study by Zilki and Resende (2021). According to the authors, those SO with psychopathy significantly present less anxiety, irritability, sadness, dysphoria, loneliness or helplessness, vulnerability to mixed affects, as well as less worries about self-destructive thoughts or suicidal intentionality in the Rorschach test (YTVC’, p = 0.050, d = 0.9; Cblend, p = .025; SC-Comp, p < 0.001, d = 1.5). Therefore, it was also observed that in both groups the SO presented low levels of stress due to their low psychological maturity level and sensibility to see the stressors factors and subtle tensions in interpersonal relationships.
Furthermore, Seto (2008) observed that men who consume child pornography content but have never committed any offense against children are less likely to commit future sexual contact offenses involving children than men who use child pornography and who have committed sexual offenses with children. This suggests that behavior is a highly predictive factor for criminal dangerousness.
Other research has shown that SO with high criminal recidivism are younger than SO with low recidivism. These people often start their criminal lives at a young age and get easily involved in socioeducational measures to fulfill their illegal acts. Therefore, it is only after the age of 40 that there is a considerable decrease in their criminal behavior (Huss, 2011).
Accurate perceptions of the dangerousness level of a person, based in the severity of the punishment applied, seem to be connected to those people who victimize strangers/no relatives, who used a gun or caused injuries (Amirault & Beauregard, 2014), and who had younger children of the same gender as a victim (Thompson et al., 2020). In turn, it was not found meaningful association of dangerousness with age, gender, marital status, years of schooling, or a profession itself (Moreira et al., 2018).
A justification to assess the crime dangerousness is to facilitate clinical decison makers, helping in the more appropriate medical and psychological interventions such as to facilitate legal decisions, helping judges understand the case in the moment of a verdict. Stinson and Becker (2016) point out that the way the risks are typically determined involves the use of assessment instruments to preview the risk by a forensic population.
Considering the above, the present exploratory study had the general objective of investigating the criminal dangerousness and personality characteristics of SO. The specific objectives were 1. to verify whether the SO who had abused children (G1) and those who had abused adolescents and adults (G2) presented differences in the criminal profile, the level of dangerousness, and personality characteristics and 2. to explore the relationship between dangerousness and personality characteristics and criminal profile in the SO.
Method
Participants
Sixty-nine SO who had abused children and adolescents participated in this study. They were randomly selected and subdivided into two groups: the first group (G1) was composed of 41 SO with exclusively child victims, and the second (G2) consisted of 28 SO with adolescent and adult victims.
The sociodemographic characteristics of the total sample (n = 69) were mean age of 31.44 years (SD = 10.78), with a minimum age of 18 and a maximum of 65 years; low education, 73.7% (n = 28), with less than four years of schooling. Regarding race/skin color, 40.0% (n = 22) of the participants declared themselves to be white, 40.0% (n = 22), mixed race and 20.0% (n = 11), black. Considering marital status, 42.0% (n = 29) were married, and 46.4% (n = 32) had children, with a mean of 2.47 children (SD = 1.66; min.: 1, max.: 9). A total of 82.6% (n = 57) of the participants performed manual work (especially in civil construction).
Regarding the inclusion criteria, the participant needed to: 1. have been convicted of sexual crimes; 2. have victimized a child and/or adolescent; and 3. have been imprisoned in a closed regime. Subsequently, the exclusion criteria were: 1. not completing the consent form; 2. performing insufficiently in the tests to provide interpretively reliable information; and 3. having moved from the closed to the semi-open or open regime during the data collection period of the study. During the research, three inmates were excluded: two due to being unwilling to respond to all data collection instruments and one due to refusing to sign the consent form.
Instruments
• Criminal Prosecution Information Collection Protocol: to collect sociodemographic data (age, education, marital status, race/color, whether they had children) and criminal profile (number of victims: victims of sexual crime; number of prosecutions; number of prosecutions for sexual violence; length of abuse; sexual crime recidivism; length of sentence; type of crime; sex of the victims: male or female; relationship to the victim: relative, neighbor, friend, or without any proximity to the victim; crime scene: abuser’s house, victim’s house, vacant lot, or other locations; whether the victim died; prison escape; participation in riot).
• Dangerousness: this variable was created using the criminal profile variables, measured through the Criminal Prosecution Information Collection Protocol. After collecting the information, principal component factor analysis (PCFA) was performed, which identified four variables to compose the dangerousness: 1. number of prosecutions; 2. number of prosecutions for sexual crimes; 3. number of victims, and 4. recidivism.
• Rorschach in the performance assessment system (R-PAS) (Meyer et al., 2017): test used to assess the personality of the SO. The Rorschach, as conceived in the R-PAS, consists of an activity that allows the examiner to observe and assess “personality in action”, quantifying and documenting important behaviors while the examinee describes what their ten inkblots could be. Through this information, the examiner can assess the examinee’s adaptability, coping style in adverse situations, underlying attitudes and concerns, and dispositions to think, feel, and act in a certain way.
In this study, the 60 variables distributed in interpretive data were used and related to the five domains of the test:
Observations (Pr, Pu, and CT): these variables describe how the examinee handled the Rorschach during its application and evoke typical behavioral aspects that they use in solving everyday problems.
Engagement and cognitive processing (complexity, R, F%, blend, Sy, MC, MC-PPD, M/MC, [CF + C]/SumC, W%, Dd%, SI, IntCont, Vg%, V , FD, R8910%, WSumC, and Mp/[Ma + Mp]): they involve the level of sophistication of the cognitive processing, the flexibility to deal with life’s demands and respond to challenges, the availability of efficient psychological resources (ideational and affective), and the ability to adapt.
Perception and thinking problems (EII-3, TP-Comp, WSumCog, FQ-%, WD-%, FQo%, P and FQu%): they evaluate the presence of severe thought disorders or ineffective and immature idiosyncratic thoughts, or flaws in judgment, as well as misinterpretations or unconventional interpretations of reality, which imply dysfunctional behaviors and failures to adapt.
Stress and distress (Y, m, MOR, SC-Comp, PPD, YTVC’, CBlend, C’, V and CritCont%): those variables provide the identification of potential and present psychological distress. Feelings of helplessness, insecurity, devaluation, and self-criticism are revealed by the variables of this domain.
Self- and others’ representation (ODL%, SR, MAP/MAHP, PHR/GPHR, M-, AGC, V-Comp, H, COP, MAH, SumH, NPH/SumH, r, p/[a + p], AGM, T, PER and An): they translate aspects related to self-image, self-esteem, and self-perception, as well as alluding to the way contacts and interpersonal interactions are established.
Procedures
For the development of the research, initially the project was approved by the Center of Excellence for Penal Execution, so that the data collection was carried out in the largest prison unit in the state of Goiás. Then, the research project was approved by the prison institution and the Research Ethics Committee, under the Certificate of Presentation for Ethical Appreciation (Certificado de Apresentação para Apreciação Ética [CAAE]) under No. 0156.0.168.000-11. After approval by the Ethics Committee, a new contact was made with the prison unit to start research into criminal cases and the application of psychological tests with the inmates. The procedures for data collection followed the steps below:
Analysis of criminal cases in the penitentiary registry to identify and collect the life history of each inmate, in view of the inclusion and exclusion criteria of the research, being selected criminal cases of inmate who had committed only sexual crimes.
After the random selection of the inmates who could participate in the study, potential participants were individually called, invited to participate in the research and sign the Informed Consent Form (ICF) in two copies: one for the participant and the other for the researcher. In this phase, only two inmates chose not to participate in the study.
Subsequently, the Roschach test was applied in a room at the prison institution, ensuring the participant's privacy during the application, which takes about 70 minutes.
The material collected was physically and electronically stored, so that all Rorschach tests protocols and information collected from criminal proceedings were registered with codes, thus ensuring confidentiality in the identification of participants.
Data analysis
All the R-PAS protocols (n = 69) were coded by the authors of this article, and 30% of them were randomly selected and sent to be coded by two judges that were experts in the test and blinded to the research objectives, to calculate the analysis of agreement between evaluators, using the intraclass correlation coefficient (ICC). The mean value of the ICC was 0.86, with a median of 0.92, ranging from 0.60 to 1.00. These values were considered to be between good and excellent, indicating evidence of reliability regarding the classification of the responses according to the R-PAS references.
Next, all the Rorschach protocols were entered into a database for basic research, located on the official R-PAS website.2 Subsequently, the data from this database and those referring to the sociodemographic and criminal profile were entered into the Statistical Package for the Social Sciences (SPSS) software, version 24.0, to perform the statistical analyses.
The Kolmogorov-Smirnov (KS) normality test was carried out, with Lilliefors correction for the 60 Rorschach variables analyzed. It was found that, in one of them (1.58%; Pu), it was not possible to perform the calculation due to zero prevalence, and only 12 (19.05%) had p > 0.05 in the KS test (F% p = 0.20; MC p = 0.06; EII-3 p = 0.05; TP-Comp p = 0.20; FQ-% p = 0.07; WD-% p = 0.08; FQo% p = 0.20; SC-Comp p = 0.20; V-Comp p = 0.20; W% p = 0.05; FQu% p = 0.07), indicating the normality of these variables. Accordingly, most of them (n = 50; 79.37%) showed deviations from normal distribution.
Subsequently, the sociodemographic data and criminal profile data from the total sample (n = 69) were investigated, and these aspects were compared between participants in G1 and G2. For the creation of the “dangerousness” variable, the PCFA was performed, using the criminal profile variables.
Of the total variables in this block of the criminal profile, six were eligible for PCFA: “number of prosecutions”, “number of prosecutions for sexual violence”, “length of sentence”, “number of victims”, “length of abuse” and “recurrence of sexual violence”. The variables “escape”, “riot”, “death” and “type of victim” were not included, due to the high number of missing data, and because they were considered qualitative in the present study (yes or no for the first three variables). Missing data are considered detrimental because they decrease the sample size and impair or “falsify” correlations.
The KS test, with Lilliefors correction, was then used to verify the normality of each variable separately. It was found that, at a significance level of 0.05, all the variables were considered to have a non-normal distribution. This was followed by the calculation of Bartlett’s test of sphericity and the Kaiser-Meyer-Olkin (KMO) statistic to verify the adequacy of the data for factor analysis. The KMO value, which was 0.60, and the p of Bartlett’s statistic indicated that the data were acceptable for factor analysis (X2: 130.84; p < 0.001) (Hutcheson & Sofroniou, 2009).
The magnitude of the correlations between the variables analyzed was then verified, with the data showing strong and positive correlations between the number of victims and the number of prosecutions for sexual violence (r = 0.63), the number of victims and sexual violence recidivism (r = 0.67), moderate and positive correlations between the number of prosecutions and the number of prosecutions for sexual violence (r = 0.56), the number of victims and the total length of the sentence (r = 0.53), the number of prosecutions for sexual violence and sexual violence recidivism (r = 0.44), and a weak correlation between the total length of the sentence and sexual violence recidivism (r = 0.33).
There were also very low correlations between the variable “length of abuse” and the other variables (p < 0.30); therefore, this variable was excluded, and the PCFA process was restarted. After excluding the variable “length of abuse”, the KMO value was 0.60, and Bartlett’s statistics (X2: 128.77; p < 0.001) again indicated that the data were acceptable for performing the PCFA (Hutcheson & Sofroniou, 2009). In the present analysis, there were no extreme values, suggesting that there were no problems in relation to commonalities.
The analysis showed the presence of two principal components (data variability: 1: 53%; and 2: 74.58%). Component 1 (Cronbach’s alpha 0.09, low internal consistency) consisted of three variables, all of which were strongly and positively correlated (above 0.60): 1. total length of the sentence, 2. number of victims, and 3. sexual violence recidivism. Component 2 (Cronbach’s alpha 0.65) consisted of two variables, both strongly and positively related: 1. total number of prosecutions and 2. total number of sexual prosecutions. Finally, due to the low value of the component consistency indicators, a variable that could contribute to this low consistency was removed: “length of sentence”. Accordingly, a new PCFA was run with four variables: 1. number of prosecutions, 2. number of prosecutions for sexual crimes, 3. number of victims, and 4. recidivism.
Accordingly, the PCFA was driven by four variables from a sample of 69 SO. More than half of the variables showed a correlation > 0.30, indicating suitability for factor analysis. The KMO measure verified the sampling adequacy for the analysis (KMO = 0.63), and Bartlett’s test of sphericity (chi-square = 105.84, p < 0.001) indicated that the correlations between the items were sufficient to carry out the analysis.
The investigation of the communalities indicated the absence of extreme values, suggesting that there were no problems in this regard. The initial analysis showed that one component complied with the Kaiser criterion, with a value (eigenvalue) greater than one, and explained 61.60% of the data variance. The scree plot showed that one principal component was positioned before the inflection. Considering the sample size and the convergence between the scree plot and the Kaiser criterion, this was the only component that remained in the final analysis. The components matrix showed that all items correlated positively and strongly, with Cronbach’s alpha coefficients for the main component being 0.75. After the identification of the PCFA, it was correlated with the variables of the Rorschach test and the development index (DI).
After identifying “dangerousness” from the PCFA, the analyses were performed through descriptive and comparative statistics, using the SPSS software. Qualitative variables were presented as absolute and relative frequencies and quantitative variables, as the mean, with a confidence interval of 95% of the mean, standard deviation, minimum and maximum. There was a statistical difference between G1 and G2: for the qualitative variables, Pearson’s chi-square test for tendency or Fisher’s exact test was performed, and for the quantitative variables, Student’s t-test was performed. In addition, the effect sizes between the groups were verified for the quantitative variables using Cohen’s d and for the qualitative variables, using Cramér’s V.
The effect size was classified as small (d = 0.20 to 0.49), medium (d = 0.50 to 0.79) or large (d > 0.80), based on the reference values suggested by Cohen (1988). For Cramér’s V, the effect was classified as small (V = 0.10 to 0.20), medium (V = 0.21 to 0.60), and large (V = 0.61 to 1.0), using the values suggested by Rea and Parker (1992) as the reference. Finally, the relationship between the criminal profile variables and the correlation with the “dangerousness” variable and the Rorschach test variables was verified, using Pearson’s correlation. In all analyses, p values < 0.05 were considered statistically significant.
Results
The criminal profile characteristics of the total study sample (n = 69) are presented in Table 1. In general, approximately 77% of the participants had a single prosecution, almost all of them being convicted of a sexual crime (87%). With regard to recidivism, 68% (n = 47) did not recommit the crime, and the remainder recommitted a sexual crime between one and five times. Considering the number of victims, 71% abused one victim, and 29%, between two and five victims. Among the victims, 59% were children, and 41% were adolescents. Some SO who had abused adolescents had also abused adults, which were considered only because the inclusion criteria would include SO who abused adolescents. With regard to the age of the victims, the mean age was 15 years (SD = 7.49).
Table 1 Descriptive and comparative statistics of the quantitative variables of the criminal profile in G1 and G2
| Variáveis | N | M | IC 95% | DP | Mín. | Máx. | t1 | p | d2 | |
|---|---|---|---|---|---|---|---|---|---|---|
| Age | TG | 69 | 31.44 | 29.05-34.08 | 10.78 | 18 | 65 | 5.39 | ||
| G1 | 41 | 36.32 | 33.13-39.75 | 10.89 | 18 | 65 | < 0.001 | 1.400 | ||
| G2 | 28 | 24.32 | 22.54-26.46 | 5.32 | 18 | 40 | ||||
| No. of victims | TG | 69 | 1.48 | 1.28-1.71 | 0.91 | 1 | 5 | |||
| G1 | 41 | 1.37 | 1.15-1.62 | 0.79 | 1 | 5 | -1.24 | 0.221 | 0.288 | |
| G2 | 28 | 1.64 | 1.29-2.06 | 1.06 | 1 | 5 | ||||
| No. of prosecutions | TG | 69 | 1.46 | 1.13-1.33 | 0.96 | 1 | 5 | |||
| G1 | 41 | 1.27 | 1.07-1.51 | 0.74 | 1 | 4 | -2.09 | 0.041 | 0.490 | |
| G2 | 28 | 1.75 | 1.35-2.19 | 1.17 | 1 | 5 | ||||
| No. of prosecutions for sexual violence | TG | 69 | 1.17 | 1.05-1.21 | 0.51 | 1 | 4 | |||
| G1 | 41 | 1.07 | 1.00-1.16 | 0.26 | 1 | 2 | -2.02 | 0.048 | 0.461 | |
| G2 | 28 | 1.32 | 1.08-1.61 | 0.72 | 1 | 4 | ||||
| Length of abuse* | TG | 69 | 7.87 | 4.25-11.98 | 16.17 | 0 | 72 | 3.62 | < 0.001 | 0.969 |
| G1 | 41 | 13.24 | 7.45-19.70 | 19.29 | 0 | 72 | ||||
| G2 | 28 | 0.01 | 0.00-0.03 | 0.04 | 0 | 0 | ||||
| Recidivism | TG | 69 | 0.42 | 0.26-0.59 | 0.71 | 0 | 3 | |||
| G1 | 41 | 0.34 | 0.16; 0.59 | 0.69 | 0 | 3 | -1.11 | 0.271 | 0.279 | |
| G2 | 28 | 0.54 | 0.28; 0.83 | 0.74 | 0 | 3 | ||||
| Length of sentence | TG | 69 | 22.42 | 18.09-28.30 | 22.52 | 4 | 180 | |||
| G1 | 40 | 16.04 | 13.75-18.63 | 7.87 | 4 | 46 | -2.95 | 0.018 | 0.665 | |
| G2 | 28 | 31.53 | 22.30-44.39 | 31.97 | 12 | 180 | ||||
| TG | 69 | 0.02 | -0.22-0.26 | 1.01 | -0.59 | 4.27 | ||||
| Dangerousness | G1 | 41 | -0.18 | -0.41-0.05 | 0.74 | -0.59 | 2.77 | -2.04 | 0.045 | 0.470 |
| G2 | 28 | 0.31 | -0.18-0.80 | 1.27 | -0.59 | 4.27 | ||||
Note. M: mean; SD: standard deviation; 95% CI: 95% confidence interval;
1 Student’s t-test for independent samples;
2 Pearson’s chi-square or Fisher’s exact test;
3 Cohen’s d;
*this mean was only considered with G1. TG = Total group. Results in bold = statistically significant results, referring to the analyzes between groups G1 and G2 with p < 0.05.
Regarding the sex of the victims, 93% (n = 64) were female. Concerning the relationship of the aggressor to the victim, 42% (n = 29) were considered relatives of the victim, and 16% (n = 11) were neighbors or friends. Regarding the place of the crime, 42% (n = 29) of the abuses took place in the victim’s house, 23% (n = 16) in the SO’s house, and 21.7% (n = 15) in a vacant lot or a scrubland. In relation to the “death of the victim” variable, 4.3% (n = 3) of the cases resulted in the death of the victim, however, this variable presented four cases of missing data. A total of 46% (n = 30) had escaped from the prison unit. In the “participated in a riot” variable, 26 cases of missing data were identified, with only 2.3% (n = 1) reported. The mean length of sentence for the total group was 22 years (SD = 22.52).
Table 2 presents the descriptive and comparative statistics of the results referring to the characteristics of the criminal profile of the two groups. These data are related to the first specific objective: to verify whether SO who had abused children (G1) and those who had abused adolescents and adults (G2) would present differences in the criminal profile. Therefore, observing the statistical significance and the size of the effect through Cramér’s V, the results showed that the G1 participants presented a greater number of prosecutions (between two and five) (p = 0.042/ Cramér’s V = 0.245) than the G2 participants, with a medium effect size. Regarding the type of crime, G1 was significantly more prosecuted exclusively for sexual crimes, and G2 for sexual crimes and others (p < 0.001/ Cramér’s V = 0.612), with a large effect size.
Table 2 Descriptive and comparative statistics of the criminal profile by group
| Variables | G1 (n = 41) | G2 (n = 28) | p2 | Cramér’s V |
|---|---|---|---|---|
| Number of prosecutions | ||||
| One | 35 (85.4) | 18 (64.3) | 0.042 | 0.245 |
| Two to five | 6 (14.6) | 10 (35.7) | ||
| Number of prosecutions for sexual violence | ||||
| One | 38 (92.7) | 22 (78.6) | 0.144 | 0.206 |
| Two to four | 3 (7.3) | 6 (21.4) | ||
| Type of crime | ||||
| Only sexual | 34 (82.9) | 6 (21.4) | < 0.001 | 0.612 |
| Sexual and others | 7 (17.1) | 22 (78.6) | ||
| Recidivism | ||||
| Zero | 31 (75.6) | 16 (57.1) | 0.208 | 0.215 |
| One | 7 (17.1) | 10 (35.7) | ||
| Two to three | 3 (7.3) | 2 (7.1) | ||
| Total number of victims | ||||
| One | 31 (75.6) | 18 (64.3) | 0.309 | 0.123 |
| Two to five | 10 (24.4) | 10 (35.7) | ||
| Gender of the victims1 | ||||
| Male | 5 (12.2) | - | 0.075 | 0.231 |
| Female | 36 (87.8) | 28 (100.0) | ||
| Relationship with the victim1 | ||||
| None | 7 (17.1) | 22 (78.6) | < 0.001 | 0.677 |
| Related | 29 (68.3) | 1 (3.6) | ||
| Neighbor/friend | 6 (14.6) | 5 (17.9) | ||
| Place of crime1 | ||||
| Victim’s house | 19 (46.3) | 10 (35.7) | 0.014 | 0.387 |
| SO’s house | 13 (31.7) | 3 (10.7) | ||
| Vacant lot/scrubland | 4 (9.8) | 11 (39.3) | ||
| Other | 5 (12.2) | 4 (14.3) | ||
| Death of the victim1 | ||||
| No | 40 (97.6) | 26 (92.9) | 0.562 | 0.113 |
| Yes | 1 (2.4) | 2 (7.1) | ||
| Prison escape | ||||
| No | 27 (73.0) | 8 (28.6) | < 0.001 | 0.441 |
| Yes | 10 (27.0) | 20 (71.4) | ||
Note.
1 Considering the first victim;
2 Pearson’s chi-square or Fisher’s exact test. Data in bold = statistically significant results, referring to the analyzes between groups G1 and G2 with p < 0.05.
Regarding the proximity of the SO with the victim, G1 chose significantly more victims to whom they were close, such as relatives, neighbors, and friends (p < 0.001/Cramér’s V = 0.677) than G2, whose victims without any closeness predominated, with a large effect size. Considering the place chosen to carry out the sexual violence, G1 preferentially chose the victim’s house and his own to carry out the crime; however, G2 tended to choose vacant lots or a scrubland (p = 0.014/Cramér’s V = 0.387), with a medium effect size. Regarding escape from the prison unit, G1 had significantly fewer escapes compared to G2 (p < 0.001/Cramér’s V = 0.441), with a medium effect size.
Also considering the first specific objective regarding the differences in the criminal profile, the “age”, “number of prosecutions”, “number of prosecutions for sexual violence”, “length of abuse”, and “length of sentence” variables were found to be statistically different between the groups. The effect sizes calculated for these variables, using Cohen’s d, ranged from small to large.
Concerning age, G1 was significantly older (p = < 0.001/d = 1.400). Regarding the number of prosecutions, G2 had more criminal prosecutions (p = 0.041/d = 0.490), with an effect size considered small. In relation to the number of prosecutions for sexual violence, G2 had significantly more prosecutions (p = 0.048/d = 0.461), with a small effect size. Considering the length of the abuse, only G1 continued the abuse (p < 0.001/d = 0.969), with a large effect size. However, in relation to the length of sentence, G2 received longer sentences (p = 0.018/d = 0.665), with the effect size considered medium.
The second specific objective was to verify whether the SO who had abused children (G1) and those who had abused adolescents and adults (G2) would present differences in the level of dangerousness. The analyzed data indicated a higher level of dangerousness among the G2 participants. The mean of the PCFA named “dangerousness” was -0.18 for G1 and 0.31 for G2 (p = 0.045; d = 0.470), with an effect size considered small, however, showing a statistically significant difference between the groups.
In view of the third specific objective, which was to verify whether the SO who had abused children (G1) and those who had abused adolescents and adults (G2) show differences in the personality characteristics, see Table 3. This table shows results with statistically significant differences, referring to the personality characteristics of G1 and G2, evidenced through the R-PAS variables, with the effect size, according to Cohen’s d, ranging from small to large.
Table 3 Rorschach test variables with significant differences between groups and their respective effect sizes
| Variables | M | 95% CI | SD | Min. | Max. | t1 | p | d2 | |
|---|---|---|---|---|---|---|---|---|---|
| AGC | G1 | 3.58 | 2.92; 4.29 | 2.26 | 0 | 9 | 3.272 | 0.002 | 0.838 |
| G2 | 1.92 | 1.30-2.58 | 1.71 | 0 | 6 | ||||
| PPD | G1 | 8.41 | 7.03; 9.79 | 4.56 | 0 | 24 | 2.130 | 0.037 | 0.520 |
| G2 | 6.00 | 4.27; 7.88 | 4.70 | 0 | 20 | ||||
| YTVC’ | G1 | 3.14 | 2.39-4.00 | 2.56 | 0 | 13 | 2.685 | 0.009 | 0.675 |
| G2 | 1.65 | 1.00-2.36 | 1.78 | 0 | 7 | ||||
| AGM | G1 | 0.80 | 0.51- 1.13 | 0.98 | 0 | 4 | 1.887 | 0.050 | 0.474 |
| G2 | 0.39 | 0.14- 0.70 | 0.73 | 0 | 3 |
Note. M: mean; SD: standard deviation; 95% CI: 95% confidence interval;
1 Student’s t-test for independent samples;
2 Cohen’s d.Data in bold = statistically significant results, referring to the analyzes between groups G1 and G2 with p < 0.05.
The results indicated more aggressive ideations, with a tendency to reflect an identification with power, danger, and threat (AGM), as well as a greater predisposition to present more sudden aggressive actions due to a lack of behavioral control and an impulsive and less rational character (AGC) in G1, with a large (AGC, p = 0.002/d = 0.838) and small (AGM, p = 0.051/d = 0.474) effect sizes for these personality characteristics.
Despite the aggressiveness, SO against children (G1) also revealed more manifestations of emotional distress than SO against adolescents and adults (G2). This suggests that G1 had more self-criticism in relation to his actions, that is, they experienced more subjectively felt stress (PPD, p = 0.037/d = 0.520) and emotional stress (YTVC’, p = 0.009/d = 0.675), with a medium effect size, when compared with the SO against adolescents and adults (G2).
Finally, Table 4 presents the relationship between the PCFA named “dangerousness” and the variables of the Rorschach test and criminal profile, in response to the fourth specific objective of this study. As it can be seen, the PCFA “dangerousness” presented a weak but significant correlation with three variables of the Rorschach test (R8910% r = 0.28, p = 0.027; Mp/[Ma + Mp] r = 0.27, p = 0.029 and IntCont r = 0.28, p = 0.027), two of the criminal profile variables (escape from prison r = 0.38, p = 0.002 and total length of sentence r = 0.36, p = 0.004) and age (r = -0.24, p = 0.050).
Table 4 Relationship between dangerousness and criminal profile variables and the Rorschach test
| Variables1 | r2 | p-value | |
|---|---|---|---|
| Rorschach | R8910% | 0.276* | 0.027 |
| Mp/(Ma + Mp) | 0.273* | 0.029 | |
| IntCont | 0.277* | 0.027 | |
| Criminal profile | Prison escape | 0.377** | 0.002 |
| Total length of sentence | 0.355** | 0.004 | |
| Age | -0.239 | 0.050 | |
Note.
*correlation significant at 0.05 level;
**correlation significant at 0.01 level; 1 variables of the Rorschach test and the criminal profile; 2 Pearson’s correlation.
These data indicate that higher levels of dangerousness equated to a greater predisposition to use intellectualization (IntCont), greater receptivity to exciting stimuli, with a more superficial processing of reality (R8910%), and greater predisposition to expecting people to fulfill their demands and fantasies (Mp/[Ma + Mp]), as well as being associated with a predisposition for more escapes, having longer sentences, and being younger.
Discussion
The present exploratory study had the general objective of investigating the criminal dangerousness and personality characteristics of SO. Similarities were found regarding the criminal characteristics of the SO in the present study with those already observed in the literature. That is, the SO of the prison investigated were between 30 and 40 years old on average, and most of them had been convicted of a crime of sexual violence. Furthermore, they tended to be people close to his victims, such as fathers, stepfathers, uncles, and neighbors, similarly to the SO from other regions of Brazil. They also preferentially abused children in the victim’s house or his own (Soares et al., 2016).
For criminalists, criminal dangerousness results from the practice of crime and is based on the danger of repeating the crime (Seto, 2008; Slaibi & Gomes, 2014). Accordingly, the SO who had abused adolescents and adults (G2) presented more danger to society in general, not only because they revealed more criminal versatility, whether for sexual or non-sexual crimes, which consequently is related to an increase in the number of prosecutions, but also because they tended to be related to a greater number of victims, criminal recidivism, and less demonstration of underlying psychological distress.
In view of the psychological characteristics, the SO against children (G1) revealed more aggressive traits, either through a sudden and impulsive lack of behavior control (AGC), such as fulfilling their fantasies and desires or acting immaturely without prior reflection, or through intrusive thoughts of hostile aggressive content (AGM). Considering the international normative means of Meyer et al. (2017) for aggressive hostile content (AGM) and sudden actions of aggression due to lack of behavioral control (AGC), G1 also presented results above the reference values for aggression (AGM: M = 0.54, SD = 0.81; AGC: M = 3.05, SD = 1.93). It should be highlighted that Brazilians also make up this international normative sample. This reinforces the predisposition in G1 to actions that normally present an impulsive and less rational character (AGC), however, the data also imply the act of thinking about these aggressive actions (AGM).
It is inferred that these impulsive actions and aggressive thoughts against children generated suffering due to the desire and low ability to control in the G1 participants. The presence of significantly higher psychological distress in G1 may favor them accepting help to change their thoughts, behaviors, and feelings, in order to decrease the subjectively felt emotional stress (Meyer et al., 2017). The significantly lower psychic suffering in SO against adolescents and adults (G2) makes them more dangerous and highlights the difficulty in changing their personality, since they tend not to suffer from their criminal behavior and remain in tune with their ego (satisfied with themselves while acting in a way that is harmful to people and the community in general) (Meyer et al., 2017).
Etcheverría (2009), when comparing SO against children and adolescents with the normative sample, observed that these SO presented less aggressive ideation. However, when compared with criminals sentenced for a type of crime other than sexual crimes, the SO showed a higher level of aggressive ideation, which corroborates the findings of the present study, according to which the SO against children (G1) tended to reveal more aggressive hostile thoughts (AGM) than the SO against adolescents and adults (G2), who managed to contain their aggressive thoughts and not make them explicit. This happens because these thoughts were not considered maladjusted or disturbing by the G2 participants, who presented more dangerousness than the G1 participants.
Ryan et al. (2008), when researching pedophiles and ephebophiles using the Rorschach test, found other characteristics that are related to aggression in a certain way, including negativism, anger, and resentment (S with FQ-), which may also have allowed the SO against children (G1) to present more aggressive ideations (AGM) and sudden actions of aggression due to lack of behavioral control (AGC) than the SO against adolescents and adults (G2) and more aggressiveness than people from a non-clinical sample. Regarding the low level of stress and psychological distress (Y), these characteristics were also noted in the SO evaluated by Zilki and Resende (2021), Etcheverría (2009), and Gacono and Evans (2011), corroborating the findings of the present study, in which the SO (G1 and G2) showed less emotional distress than the average distress patterns for an international sample of R-PAS non-patients. When compared to each other, G2 showed significantly less suffering than G1.
These findings indicate that people who abuse children reveal more sudden aggressive actions and aggressive hostile thoughts. These characteristics become evident when an adult, who would be closer to their highest level of psychological maturation, starts to sexually desire a child, who is in the developmental phase. Therefore, every act or speech of a sexual nature will be aggressive for children up to 11 years old, who are not yet biopsychologically prepared to come into contact with this typical adult sexuality.
Gacono and Evans (2011) make an observation because, for the authors, a psychopath will reveal hardly any more aggressive ideations (AGM) than other people or criminals, because these thoughts can be filtered and not shared. However, sudden actions of aggression due to lack of behavioral control (AGC) are a more impulsive type of aggression, typical of people who lose control of their behavior.
Despite this, even considering the situation of deprivation of liberty of SOs sand the living with the precariousness of prison, the level of suffering and psychological stress of the SOs against children (G1) of the present research was below that of the non-clinical general population (Meyer et al., 2017). That is, people from a non-clinical sample (of non-prisoners) tended to suffer more than the SO against children (G1) of this research. In turn, it can be identified, even according to the statistical analyses presented in this study, that SO against children tended to suffer significantly more than SO against adolescents and adults (G2). In this sense, G2 showed fewer characteristics of psychological distress. These results corroborate Zilki and Resende's (2021) investigation, in which SO had low levels of stress due to a low level of psychological maturity and little sensitivity to perceive environmental stressors and more subtle tensions in the interpersonal relationships.
Apparently, part of the repertoire of the G2 participants, in order not to come into contact with suffering, is the predisposition to use intellectualization (IntCont) and the fantasy that people should fulfill their demands and desires, which helps to neutralize the effects that painful emotions could have on their cognitive processes. It is understood that this would work as a defense mechanism, so that they do not come into contact with situations that cause pain and anguish (Meyer et al., 2017).
Similar defense strategies, through distorted thinking, were also observed by Stinson and Becker (2016). According to the authors, distorted attitudes and beliefs, which favor the sexual offense, are one of the predictors for criminal recidivism. In turn, Ward (2000) proposed a cognitive model named Implicit Theories (IT) to explain the cognitive distortions present in SO. Accordingly, SO would have IT regarding the relationships between children and sex, such as: “Children are sexualized beings and provoke adults”, “I deserve to have sex when I want it”, “The world is dangerous and people are not trustworthy, having sex with children is safer”, and “Sexual intercourse with children does not harm them” (Walton et al., 2017). This would therefore favor engaging in socially reprehensible sexual behaviors with children.
The data from the present study also showed that the greater the danger, the greater the receptivity to emotionally stimulating situations (R8910%). This receptivity may not be a positive or negative aspect of a person’s personality, however, Gacono and Evans (2011) observed that it can be an aspect that harms people who have problems controlling or modulating affects, as they tend to be individuals who seek emotional experiences that predispose them to untimely and passionate emotional reactions. In addition to what the authors observed, in the case of people with high dangerousness, this predisposition to remain in environments where people more spontaneously express their affections and feelings could be a way of identifying people who are more emotionally vulnerable and possible victims.
Considering the characteristics of the criminal profile, it was observed that the greater the danger, the greater the number of escapes from the prison unit, the longer the sentences tended to be, and the younger the inmates. In the study by Stinson and Becker (2016), criminal recidivism in SO, or the dangerousness, was related to a series of aspects present in the criminal profile of the study participants. Among these aspects, there was young age. Studies have shown that SO with high criminal recidivism tend to be younger than SO with low recidivism. These people often start criminal life at a young age, and it is only after age 40 that there is a considerable decrease in their criminal acts (Huss, 2011). Therefore, the younger the age of the SO, the greater the level of danger that they can pose to victims, corroborating the findings of the present study, which indicated that young age of the SO is related to a higher level of dangerousness.
In addition to age, Stinson and Becker (2016) highlighted a history of imprisonment for sexual and non-sexual crimes, patterns of violence associated with sexually offensive behavior, antisocial characteristics, deviant sexual interests, and sexual self-regulation problems as indicative of greater danger in SO. These findings concerning the prison history of SO may be somehow related to longer sentences, as the difficulty in sexual self-regulation contributes to the SO’s recidivism, causing them to be given longer sentences. Regarding the greater number of escapes from the prison unit, this may be related to criminal versatility, with these antisocial personality characteristics being found in SO. Therefore, the data found indicate that behavior is a highly predictive factor of criminal dangerousness.
Studying the criminal dangerousness and personality characteristics of SO can contribute to the development of psychological intervention strategies, as well as specific interventions for those who present a greater danger to society. Accordingly, future studies, which can add more information about the criminal history of SO, may help with the investigation of new findings on the dangerousness and personality characteristics of SO. Future research should also analyze a sample exclusively composed of women convicted of sex crimes. The present sample was limited to male inmates.
