Risk Assessment and Dangerousness Evaluation in Forensic Psychology

When a court needs to decide whether someone should be released on parole, or when a prison system must determine the right level of supervision for an offender, one question sits at the heart of it all: how likely is this person to cause harm in the future? That question is what risk assessment in forensic psychology sets out to answer. It’s a process that blends psychological evaluation with legal considerations, and it directly shapes decisions about sentencing, parole, rehabilitation, and public safety. But predicting human behavior is inherently uncertain – and that uncertainty makes risk assessment one of the most challenging and consequential tasks in the field.

What risk assessment actually means in forensic psychology 🔗

Risk assessment in forensic psychology is the systematic process of evaluating how likely an individual is to engage in harmful, violent, or criminal behavior in the future. It is not – and this is an important distinction – a prediction that someone will be dangerous. Rather, it estimates the probability of certain behaviors occurring under certain conditions. As researchers at the Oxford Uehiro Centre for Practical Ethics note, more than 200 structured tools are now available for assessing violence risk in forensic psychiatry and criminal justice settings. These tools inform decisions on initial sentencing, parole, post-release monitoring, and rehabilitation planning.

Forensic psychologists, psychiatrists, and criminal justice professionals carry out these evaluations. They draw on criminal records, psychiatric evaluations, psychological testing, interviews, and information about the individual’s social and family background. The goal is not to label someone as “dangerous” or “safe” but to understand the nature, degree, and context of the risk they may pose.

Risk versus dangerousness: an important distinction 🔗

In everyday language, “risk” and “dangerousness” are often used interchangeably. In forensic psychology, they mean different things. Risk refers to the statistical likelihood that an individual will reoffend or engage in harmful behavior. It’s expressed in probabilistic terms – for example, a certain percentage chance of reoffending within five years.

Dangerousness, on the other hand, refers to the severity of harm that could result from an individual’s behavior. Someone might have a relatively low statistical risk of reoffending, but the potential consequences of that reoffending could be catastrophic – say, a serious violent crime. Conversely, a person could have a high likelihood of committing minor offenses without posing a threat of severe harm.

This distinction matters because it shapes how the justice system responds. A person assessed as low-risk but high-dangerousness may still warrant close supervision, while a high-risk but low-dangerousness individual might be better served by community-based rehabilitation programs. Effective forensic evaluation considers both dimensions together.

Static and dynamic risk factors 🔗

Risk assessment relies on identifying and weighing specific risk factors – characteristics or circumstances that are empirically associated with reoffending. These factors fall into two broad categories.

Static risk factors 🔗

Static risk factors are historical and unchangeable. They include things like age at first offense, gender, number of prior convictions, history of childhood abuse, and the nature of past crimes. Because they don’t change over time, static factors are useful for estimating long-term risk but offer limited guidance on whether an individual’s risk level has changed since their last assessment. Research consistently shows that static criminal history is among the most robust predictors of future reoffending, particularly for general recidivism.

Dynamic risk factors 🔗

Dynamic risk factors are those that can change. These include substance abuse, antisocial attitudes, employment status, mental health symptoms, social support networks, and impulsivity. According to a meta-analysis published in Criminal Justice and Behavior, dynamic risk factors were more strongly related to recidivism than static factors within the same risk domain for forensic outpatients. This makes them especially valuable – not just for predicting risk, but for identifying what treatment and intervention strategies should target.

Dynamic factors are further divided into stable factors (which change slowly, such as personality traits or deeply held attitudes) and acute factors (which can shift rapidly, like a recent job loss, substance relapse, or access to weapons). A prospective study from New Zealand involving over 75,000 assessments found that changes in acute risk factors were the best predictors of imminent violent recidivism, outperforming both static measures and initial risk scores.

Protective factors 🔗

Risk assessment isn’t only about what makes someone more likely to offend. Protective factors – such as strong family bonds, stable employment, engagement in treatment, and prosocial peer relationships – can buffer against reoffending. Comprehensive risk assessments account for these protective factors as well, as they are essential for developing realistic and effective risk management plans.

Approaches to conducting risk assessment 🔗

Over the decades, the methods used to assess risk have evolved significantly. There are three main approaches, each with distinct strengths and limitations.

Unstructured clinical judgment 🔗

This was the dominant method before the 1990s. A clinician would form a subjective impression of an individual’s risk based on their professional experience and expertise. While clinical experience remains valuable, research from the 1970s onward showed that unstructured clinical judgment was highly subjective, vulnerable to cognitive biases, and sometimes no better than chance at predicting future violence. This led to the development of more structured methods.

Actuarial risk assessment 🔗

Actuarial tools use statistically derived algorithms to estimate risk. They take a fixed set of risk factors – often mostly static – assign numerical values, and combine them into a total score that corresponds to a probability of reoffending. Examples include the Static-99R (widely used for sexual offense recidivism), the Violence Risk Appraisal Guide-Revised (VRAG-R), and the Level of Service Inventory (LSI).

The main advantage of actuarial tools is objectivity: they remove human judgment biases from the scoring process. A review published by the U.S. Courts noted that meta-analyses have found actuarial assessment to be approximately 13 percent more accurate than clinical judgment overall, and even more accurate for predicting violent or criminal behavior specifically. However, actuarial tools have significant limitations. They don’t account for unique individual circumstances, they provide little guidance on how to manage risk, and they can’t tell you what kind of violence to expect or under what conditions it might occur.

Structured professional judgment (SPJ) 🔗

SPJ represents something of a middle ground. It uses empirically supported guidelines – a defined set of risk factors drawn from research – but allows the assessor to exercise professional discretion in weighing those factors and forming a final judgment. Unlike actuarial tools, SPJ does not produce a mechanical score. Instead, it guides the evaluator through a systematic process that includes identifying risk factors, assessing their relevance to the individual case, and developing a risk management plan.

Commonly used SPJ tools include the Historical, Clinical, Risk Management-20 (HCR-20) for general violence risk, the Structured Assessment of Violence Risk in Youth (SAVRY) for juveniles, and the Spousal Assault Risk Assessment (SARA) for intimate partner violence. Research comparing SPJ and actuarial approaches has found that SPJ-based risk estimates perform comparably to actuarial total scores in predicting violent offending, and both significantly outperform unstructured clinical judgment.

A key advantage of SPJ is its focus on risk management, not just risk prediction. By considering dynamic factors and protective factors, SPJ evaluations can directly inform treatment planning and supervision strategies.

Key tools used in forensic risk assessment 🔗

Forensic psychologists have a wide array of instruments at their disposal. Some of the most commonly used include:

HCR-20 (Version 3): A 20-item SPJ tool that examines historical factors (e.g., past violence, substance use, personality disorder), clinical factors (e.g., current insight, violent ideation), and risk management factors (e.g., treatment compliance, stress). It’s one of the most widely validated violence risk tools in the world.

Static-99R: An actuarial tool specifically designed for adult males convicted of sexual offenses. It uses 10 static items – including age, prior offenses, and victim characteristics – to estimate probability of sexual recidivism over 5, 10, or 15 years. Validation studies in California have confirmed its predictive accuracy across diverse ethnic groups.

PCL-R (Hare Psychopathy Checklist-Revised): While not strictly a risk assessment tool, the PCL-R measures psychopathic personality traits that are strongly associated with violent and criminal behavior. It is frequently used as part of broader risk evaluation batteries.

VRAG-R (Violence Risk Appraisal Guide-Revised): An actuarial instrument that predicts violent recidivism using factors such as age, criminal history, and psychopathy scores.

LSI-R (Level of Service Inventory-Revised): Assesses both risk and needs across domains like criminal history, education/employment, family, substance use, and attitudes, making it useful for both classification and treatment planning.

The role of recidivism in risk assessment 🔗

Recidivism – the tendency of a convicted offender to reoffend – is at the core of most risk assessments. The entire framework rests on identifying who is most likely to commit further offenses and what can be done to reduce that likelihood.

Research guided by the Risk-Need-Responsivity (RNR) model, one of the most influential frameworks in correctional psychology, holds that effective interventions should match the intensity of treatment to the offender’s risk level, target the specific dynamic risk factors (or “criminogenic needs”) driving their behavior, and be delivered in a way that suits the individual’s learning style and abilities. According to a large-scale study in the Journal of Criminal Justice, antisocial peers, education and employment problems, antisocial attitudes, and substance abuse were among the dynamic factors most significantly correlated with reoffending.

By identifying these factors, risk assessment doesn’t just predict recidivism – it actively guides the design of intervention programs aimed at reducing it.

Challenges and limitations of risk assessment 🔗

Despite significant advances, risk assessment in forensic psychology faces several serious challenges.

The problem of false positives and false negatives 🔗

No risk assessment tool is perfectly accurate. A false positive occurs when someone is classified as high-risk but does not reoffend – potentially resulting in unnecessary incarceration or overly restrictive supervision. A false negative is the opposite: someone assessed as low-risk goes on to commit a serious offense. Early research found that clinicians identifying individuals as dangerous were wrong more often than they were right. While tools have improved substantially, the trade-off between false positives and false negatives remains an inherent challenge.

Racial and demographic bias 🔗

There are legitimate concerns that risk assessment tools may perpetuate or amplify existing biases in the criminal justice system. Static factors like criminal history can reflect patterns of over-policing in minority communities rather than genuine differences in offending. A study examining youth offenders found that static risk factors were a significant predictor of recidivism for White youth but not for Black youth, suggesting that these factors may function differently across racial groups. This raises important questions about fairness and equity in risk-based decision-making.

Predicting rare events 🔗

Serious violent acts are relatively rare, even among offender populations. Statistically, predicting rare events is inherently difficult – the base rate problem means that even a reasonably accurate tool will generate a high number of false positives when trying to identify low-frequency outcomes. This is particularly relevant for decisions about indefinite detention or high-security placement, where the consequences of error are severe.

The limits of individual prediction 🔗

Actuarial tools are built on group-level data. They can tell you that, among 100 people with a similar profile, a certain percentage are likely to reoffend. But they cannot tell you with certainty whether this specific person will be one of them. Translating group probabilities to individual cases remains a fundamental limitation. As researchers in the European Psychiatry journal have argued, better and more unbiased data are needed to improve how risk tools are applied at the individual level.

How risk assessment shapes offender management 🔗

Despite its limitations, risk assessment is an indispensable part of forensic decision-making. It informs key decisions at nearly every stage of the criminal justice process:

Sentencing: Risk assessments help judges understand the likelihood of future offending when determining appropriate sentences. A high-risk individual may receive a longer sentence or more restrictive conditions.

Parole and release decisions: Parole boards use risk evaluations to decide whether an offender can be safely reintegrated into the community and under what conditions. As a Swiss study published in Frontiers in Psychiatry highlighted, the way these evaluations are communicated to offenders significantly affects their engagement with the process and their motivation for change.

Treatment planning: By identifying specific criminogenic needs, risk assessments guide the design of individualized treatment programs – substance abuse therapy, cognitive-behavioral interventions, anger management, and more.

Supervision levels: Risk classifications help correctional agencies allocate resources efficiently, providing intensive supervision to those who need it most while allowing lower-risk individuals more freedom.

The evolving landscape of forensic risk assessment 🔗

Risk assessment continues to evolve. The field has moved from an era of subjective clinical impressions to one grounded in empirical evidence and structured methodology. Current best practice increasingly favors combining actuarial data with structured professional judgment – using statistical tools as a foundation while layering in individualized clinical analysis and risk management planning.

There’s also growing recognition that risk is not a fixed property of a person. It fluctuates based on circumstances, treatment, support systems, and life events. This understanding is driving a shift toward dynamic, ongoing assessment rather than one-time evaluations – monitoring how risk factors change over time and adjusting management plans accordingly.

Ultimately, forensic risk assessment sits at the intersection of science, ethics, and law. It carries enormous weight: the freedom of individuals and the safety of communities depend on its accuracy and fairness. The challenge for the field is to keep refining its tools and methods, to be transparent about their limitations, and to ensure that risk-based decisions are made as equitably and as wisely as possible.

What do you think? Should risk assessment tools carry more weight than clinical judgment in decisions about someone’s freedom – or does the human element of professional discretion remain essential? And how should we address the potential for demographic bias in these tools without compromising their predictive accuracy?

References

1. https://verdugopsych.com/violence-risk-assessments-a-guide-to-evaluating-violence-risk-in-criminal-cases/
2. https://pmc.ncbi.nlm.nih.gov/articles/PMC5408162/
3. https://www.ncbi.nlm.nih.gov/books/NBK396457/
4. https://journals.sagepub.com/doi/10.1177/0093854819826109
5. https://pubmed.ncbi.nlm.nih.gov/34928647/
6. https://www.ncbi.nlm.nih.gov/books/NBK261197/
7. https://www.uscourts.gov/sites/default/files/75_2_9_0.pdf
8. https://onlinelibrary.wiley.com/doi/full/10.1002/bsl.2698
9. https://saratso.org/index.cfm?pid=1360
10. https://www.sciencedirect.com/science/article/abs/pii/S0047235213000950
11. https://journals.sagepub.com/doi/10.1177/0306624X211022673
12. https://www.frontiersin.org/journals/psychiatry/articles/10.3389/fpsyt.2021.643096/full