Aggressiveness is a pivotal behavioral concern associated with psychopathy. Though in forensic psychiatric settings, a reliable assessment comes with diverse conceptual and methodological challenges. Psychopathy is classically defined as a combination of affective, interpersonal, and behavioral traits such as callousness, lack of empathy or remorse, manipulative tendency, impulsivity, irresponsible behavior, and antisocial conduct. In correctional settings, high levels of psychopathy often go hand in hand with high rates of aggressive phenomena in different physical and non-physical forms [1, 2].

The consistency of the association between psychopathy and aggressiveness depends largely on the psychometric measures used in assessment, the setting where assessment is applied, whether clinical, community or forensic, and lastly, if aggression tendency is evaluated by self-reporting or external ratings as clinician-rated tests or prison record. Another aspect further complicating aggressiveness’ assessment in psychopathy, is often met when assessment involves tools with different sources of information feeding its quantitative final score. The Hare Psychopathy Checklist (PCL) is widely utilised as the gold standard test for psychopathy level in forensic research and forensic expertise in the European Union EU. It involves both data from records and a structured interviews to rate objectively psychopathic traits [3, 4]. In contrast, self-report methods such as LSRP (Levenson Self-Report of Psychopathy) reflect the individual's subjective endorsement of qualities and self-perception which may represent response bias due to social desirability and lack of insight [5–7].

The definition of aggression can be a far from simplistic task. The Buss Perry Aggression Questionnaire (BPAQ) measures not only physical types of aggression but also other common nonphysical aspects as verbal, anger, and hostility. It is widely used as the golden standard psychometric tool in forensic researches for aggression tendency. All those subscales are measured through self-report questionnaires [8–10].

When subjective and objective measures are compared, discrepancies often shows-up, some individuals tend to under-estimate or over-estimate their aggressive tendencies. This disparity between tests explains, at least in part, the non-alignment between psychopathy and aggressive tendencies in sub-populations. In forensic settings, the type of differences just mentioned becomes particularly pertinent, since individuals with psychopathy traits tend to deny aggression in oneself while professionals over-rely on structured psychometric ratings and records of disciplinary measures, that sanction violent behviors [11–13].

To what degree self-report and clinician-rated psychopathy agree, and how each relates to aggression, is essential for assessment of risk, care planning, and legal decision-making [14, 15]. The difficulty to measure aggression or psychopathy is further complicated in forensic settings. This paper attempts to better understand these complexities and opens new insights on future directions.

Data collection for this study was done at Ducos Prison, Martinique, DOM Department oversees of France. It has a capacity of 730 people altogether. Among records of male subjects with a formal diagnosis of antisocial personality disorder (ASPD) who visited the psychiatry clinic during this period from Jan 2025 to June 2025. Only computerized records that included data from all three psychometric tests of interest Hare PCL, LSRP, BPAQ were selected for the analysis. We excluded records that show a comorbid borderline personality disorder or active psychosis to limit confounding effect of those comorbidity on aggression subtypes and to include only one personality disorder of antisocial type. All records included in the analysis confirmed at least one passage before a disciplinary commission for violent behavior at the prison, to further add an objective aspect of aggressiveness.

The study was authorized by the local ethics committee IRB of the University Hospital CHU de Martinique, approval 2025/045. Because the design is retrospective and relies upon pre-existing clinical data, we have no means to influence which cases are more likely to be missing data due to non-availability or loss over time. This is why sample sizes differ between correlation analyses. In particular, the sample size is n = 47 for analyses involving the PCL, LSRP et BPAQ. There were 4 more records incorporating BPAQ and LSRP but not PCL, that’s why certain analysis has been made on 51 participants as correlation between LSRP and BPAQ.

Psychometric Measures

Hare Psychopathy Checklist (PCL)

The most frequently used clinician-rated instrument for assessing psychopathy is the Hare Psychopathy Checklist, usually in its revised format (PCL-R) or some modification based on venue [2]. The PCL is a criminology instrument based on documentary review, collateral information from relatives and informants, and structured interviews with offenders. It assesses 20 items on a 3-point scale ranging from 0 = absent to 2 = present for affective, interpersonal, behavioral and antisocial aspects [2, 3].

The total score ranges from 0 to 40, with higher scores indicative of greater psychopathy level. The PCL has a two factor model; the two are interpersonal and affective traits and the second factor is the interfering deviant lifestyle strategies. Whether to adopt a three hierarchical factor or some other alternative factor model has been proposed [16, 17].

In forensic samples, the intraclass correlation for the PCL is usually very high (often 0.8) and its internal consistency is moderate to good depending on sample. The PCL shows strong and widely established predictive validity for recidivism, violence, as well as institutional disciplinary infractions [3, 4]. Some critiques persist, including possible bias in scoring as a function of case completeness and suggested adversarial allegiance conditions in carceral contexts. Since it depends on supplementary information and clinical opinions, the scoring process may offer less consistent results in settings with less strict documentation. In our study, we utilized the revised version of PCL in French language [18].

Levenson Self-Report Psychopathy Scale (LSRP)

LSRP is a 26-item self-report questionnaire developed at California Medical Facility by Levenson et al.[6] to study psychopathic traits from healthy non-institutionalized Caucasian adults approximately years old. Respondents indicate their agreement or disagreement with statements on a 4-point Likert scale (1 = strongly disagree; 2 = disagree; 3 = agree; 4 = strongly agree).

The LSRP has three scores: A primary psychopathy scale (LSRP-P) for those people who show these pathological personality characteristics generally from their earliest years, they were probably children with conduct and disruptive behaviors. The Primary LSRP sub-component correlates with low agreeableness and narcissistic traits, while Secondary LSRP sub-component correlates more strongly with disinhibition, negative emotionality, and generally antisocial attitude [7]. The secondary psychopathic trait measure changes across profiles for individual patients, showing their salient scores as distinct from other factors shared between certain persons or patient groups as in individual with intelligence and low empathy. The internal consistency of the subscales has sometimes been modest (Cronbach’s alpha in the 0.60–0.75 range), and the test is subject to socially desirable responding or poor symptoms insight [19]. A more recent “expanded” version (E-LSRP) has been developed to further enhance psychometric properties of the classical version [20, 21].

In validation studies, even with self-reported results, we see that different traits are weighted differently and ultimately were significant and differential predictors of the extent someone may cross over moral lines [6]. In our study we used the French validated version of the Levenson self-report psychopathy scales, and set itself as a credible forensic psychopathy assessment tool widely used in francophone jurisdictions [22].

Buss Perry Aggression Questionnaire (BPAQ)

The Buss Perry Aggression Questionnaire (BPAQ) is a 29-item self-report inventory designed to assess trait aggressiveness on four-factor scales; physical aggression, verbal aggression, anger, hostility. Items are rated on a 4-point Likert scale from 0 (extremely uncharacteristic) to 3. BPAQ is extensively used in both behavior research and clinical studies. Usually, the summed score is seen as the global score [23].

BPAQ shows very acceptable internal consistency (Cronbach's α is usually higher than 0.80 for the global scale, with subscales lower), as well as fair to good test–retest reliability. A four-factor model is supported by several studies, but the loadings of factors will differ in forensic populations. The BPAQ correlates significantly with both behavioral and self-reported aggression in community and clinical sample [23, 24]. We used a validated French version [25].

Descriptive data

All subjects were males having antisocial personality disorder, they had no comorbid borderline personality disorder or active psychotic episode. the mean age was 32 ± 8.4 years, none of them have university education level, all of them have passed at least once before disciplinary commission with subsequent passage in disciplinary section of the prison further confirming objective records for aggressiveness. All participants had comorbid substance use disorder.

Correlation analysis

The correlation between PCL and global score of BPAQ did not reach significance (r = 0.14; p = 0.36). In contrast, the LSRP global was significantly correlated to BPAQ score (r = 0.64; p < 0.0001) this significance extended to both primary and secondary psychopathy constructs (r = 0.51 for primary, r = 0.62 for secondary, both p < 0.0001). Strikingly the correlation between PCL and LSRP in our sample was not significant (r = 0.12, p = 0.47). (Table 1)

Comparative analysis between the strength of correlations

Comparison between the correlation coefficients LSRP global/BPAQ (r₁ = 0.64) versus Primary psychopathy/BPAQ (r₂ = 0.51) didn’t show significant difference. Similarly, comparison between the correlation coefficients LSRP/BPAQ (r₁ = 0.64) versus Secondary psychopathy/LSRP (r₂ = 0.62) didn’t reach significance either. Therefore, despite the strong correlation between aggressiveness BPAQ and Psychopathy LSRP global scores, it did not significantly differ in strength from correlation between Aggressiveness BPAQ and either constructs of Primary or Secondary psychopathy. This finding shows that both primary and secondary psychopathy dimensions contribute similarly to the overall phenomenon of aggressiveness in individuals with Antisocial personality. (Table 2)

Linear regression models

Various univariate linear regression analysis models were generated and evaluated for significance by F test and the R2 determination coefficient was calculated to explain the part of variance in the outcome that’s explained by the model. Five sets of regression models were tested, the dependent variable was Aggressiveness measured by BPAQ global score then, in subsequent sets models, the global score was replaced by the subtypes of aggressiveness, physical, verbal, anger and hostility respectively. In each set of regression models, PCL score was used as the independent variable in one model, then LSRP was used as the independent variable in another step. Regression Models did not reach significance (p > 0.05) when PCL was the independent variable while global score of BPAQ or even its four aggression dimensions were the dependent variables (Outcome). This finding put in doubt the PCL as predictive of aggressiveness, at least in our sample.

On the other hand, the regression models, when LSRP was the independent variable, were all significant (p < 0.05). In the first model, Global BPAQ (Outcome) and LSRP (Explanatory factor) was highly significant (F (1,49) = 34.41; p < 0.0001). When analyzing the four sub-dimensions of aggression as the outcome while always the LSRP as the independent variable (Explanatory factor), all four models were significant (p < 0.05). These findings show LSRP as a significant predictor of aggressiveness with good effect size of 41% for global aggressiveness, 32% for physical aggression and 34% for anger. The effect size was week for verbal aggression (14%) and for hostility (13%). (Table 3)

The present study reveals the deep complexity in interpreting the relation between psychopathy and aggression in forensic populations. Our primary outcome is a disparity between clinician-rated and self-report scales highlighting the potential conflictual impact of the assessment method on the observed outcomes. The PCL, viewed by many as the standard forensic psychometric tool in psychiatric forensic expertise, also appeared to have a weak non-significant correlation with aggression (BPAQ). In contrast, the self-report LSRP yielded a strong and statistically significant correlation that explained 41% of the total variance of global measures of aggression. This lack of overlap is also marked by the non-significant correlation between the PCL and LSRP themselves, meaning that when used in this unique carceral setting, these measures may be rather assessing different facets inherent to the psychopathy construct and reflecting conceptual difference in forensic expertise assessment.

The large discrepancy between clinician-rated Psychopathy and self-report aggression can be viewed from theoretical and methodological perspectives. The PCL is a retrospective, behavior-based rating model highly dependent on extensive file search and recorded antisocial history. Incarcerated samples may have been subjected to less accurate record keeping, and thus PCL scores could be indicative of a more lagged or historical conception of psychopathy that is unrelated to present-day self-perceived aggression. In contrast, the BPAQ assesses trait aggression as an internally based and pervasive throughout patient’s life, which might be more contemporary and less dependent on the history of reported data. By contrast, the highly significant correlation between LSRP and BPAQ is also probably, at least in part, ascribable to common method variance bias, a serious confounding effect whenever some percentage of variance is not derived by constructs themselves, but rather by the methodology of their assessment [26]. With both the predictor and outcome coming from the same self-report informant, correlations may be inflated by shared method variance such as consistent response patterns and ability to perceive his own psychopathology under-estimated, whether relevant to aggressiveness or psychopathy, due to social desirability bias [27].

Our subtler results at the LSRP subscale level provide additional explanation. Primary psychopathy representing affective-interpersonal factor (egocentrism and lack of empathy) and secondary psychopathy representing the impulsive-antisocial construct manifesting through the antisocial maladaptive coping strategies, both dimensions of psychopathy were both significantly correlated to global aggression. This may indicate that within this high-risk carcerated ASPD sample, the etiological pathway to aggressiveness are diverse, consistent with recent facet-based perspectives on psychopathy [28]. The lack of emotional attachment and manipulative attitude of primary psychopathy could make it easier for proactive aggression to occur, in other terms, provocative aggression, whereas impulsivity and negative affect associated with secondary psychopathy often are strongly associated with reactive aggression. The global LSRP score, reflecting a combination of these facets, is a strong predictor of general self-reported aggression but likely masks these distinct drives behind aggression. Our findings are in agreement with multiple studies in literature emphasizing the predictive ability of LSRP regarding aggressive and violent behaviors [28, 29]. The regression analyses provide further support to this model. The LSRP emerged as a strong correlate predictive factor for both physical aggression and anger, two constructs that are core to the behavioral dysregulation imminently present in ASPD. Its weaker association with verbal aggression and hostility may suggest that these forms are more context-sensitive [28, 29].

It is particularly notable and worthy mentioning, that the PCL failed to demonstrate any predictive validity across the aggression dimensions. It challenges the view that high levels of PCL automatically result in higher self-reporting of aggressiveness. This unpredicted rand striking result may suggest that the predictive ability of PCL for aggressiveness is contradictory, depending on the context, the setting and the availability of registered data.

The present study aligns with some data from literature. A study with a large sample of USA federal offenders (1,181 male and 435 female) showed very weak correlation between Psychopathy check list and perceived aggressiveness assessed by Buss Perry aggression questionnaire and points to a potential methodological disparity between objectively-evaluated and subjectively-perceived construct [30].Another study carried out on 129 prisoners found only the interpersonal manipulation construct of psychopathy is correlated to reactive type of aggression rather than proactive provocative aggression [31].

This finding raises the importance of differentiation between the two subtypes of aggressive behavior; one proactive where the person initiates violent act without clear provocation by others, and a reactive subtype of aggression in response to provocation of others, both subtypes can be understood based on primary and secondary psychopathy dimensions assessed by LSRP but on the other hand, aggressivity assessed by BPAQ doesn’t distinguish such category-difference subtility. It’s a limitation in our study that the type of aggression was not considered in the assessement.

These results have implications for psychiatric legal practice and clinical treatment. The message we deliver is that relying too much on any one assessment method, whether objective clinician-rated as PCL or the more subjective self-reporting LSRP, creates an incomplete and possibly misguided outcome for the forensic expertise. An integrated risk calculation combining these different sources of data is required to constitute a complete risk profile in clinical practice. A deviation where self-reported aggression and psychopathy are high, yet clinician-rated psychopathy is moderate, may point to a person who has substantial internal anger but less widespread documented history of antisocial behavior. This incongruity between self-report and clinician-rated psychopathy has been extensively examined by researchers who questioned the consistency, between both psychometric tools, to evaluate the same psychopathy complex construct [32]. The discrepancy observed between the self-rated and clinician-rated psychologic measures have been emphasized and discussed with contradictory results in quite different pathologies, that also inflate the bias of the methodological approach on either side [33, 34].

Though, our findings should be seen from therapeutic perspectives, Our study revealed LSRP as a significant predictor of aggressiveness in its integrality and across it’s all sub-dimensions. Antisocial personality disorder ASPD was generally considered to be untreatable. Recent treatment approaches such as Metallization based therapy MBT for affective problems [35, 36] and Schema Therapy ST for aggression [37, 38], have shown promise in treating the central cognitive and affective deficits underlying aggressive behavior. MBT is designed to enhance mentalizing, the ability to understand their own stream of thoughts and that of others, which has been demonstrated in controlled trials to result in diminished anger and aggressive behaviors among individuals with ASPD. These improvements support the notion that a more subtle understanding is a stepping-stone to making individualized interventions more successful.

Managing this patient population also represents special challenges for physicians. Studies have indicated that therapists working with patients with ASPD commonly experience emotional reactions to these individuals, including fear, distancing, and frustration. These feelings can interfere both in the therapeutic alliance and influence clinical decision-making. Attention to these countertransference reactions is important for managing the dilemmas of treatment to build a good therapeutic relationship.

Limitations and Future Research

There are several limitations to the current study. Due to the retrospective nature and the sample size, statistical power is consecutively limited, and generalization may be restricted. The convenience sampling based on the existing clinical data may cause a selection bias. We were also missing significant confounders such as intelligence. Longitudinal, prospective research with larger and more heterogeneous forensic samples should be conducted in the future. Studies should be rely on multi-traits, multi-measures matrices that combine self-report and sensitive clinical ratings, as well as behavioral measures of aggression as institutional infractions registries. Investigating the ability of these diverse tools of assessment to predict different types of recidivism would be of great importance for forensic health policies.

The evaluation of how psychopathy contributes to aggression in a correctional setting is a sophisticated task, significantly impacted by preferences towards psychological tools. Our findings indicate that self-report (LSRP) and clinician-rated (PCL) measures of psychopathy are not equivalent and may be measuring different aspects of the same coin. This discrepancy is not just a statistical artifact, but it is rather a clear clinical pattern and reflects the complexity of both constructs, psychopathy and aggression. In forensic psychiatric expertise, the recognition of strength and weakness of the methodological assessment is critical. An integrative, multi-method model is more adapted strategy for evaluation, since it considers the complexity of the constructs and pertaining decisions based on the assessment.