Criminal risk assessment has been intensively investigated over the last decades. Therefore, the four generations of risk assessment instruments have already been developed (Blomberg et al., 2010) starting from unstructured clinical judgment (first generation), via actuarial risk assessment (second generation), and leading to structured clinical judgment (third generation) and assessment-based risk management (fourth generation). The HCR-20 violence risk assessment scheme (Webster et al., 2007) is an instrument of the third generation, which recognizes the importance of both static and dynamic risk management factors as well as contains the elements of prediction, a specific assessment procedure, and risk communication. The HCR-20 can be applied in different settings: forensic, correctional, or community. Moreover, the research has shown (Dahle, 2006; Gray et al., 2008) that the HCR-20 is reliable for assessing the risk of general and non-violent criminal behavior.
In Lithuania, risk assessment is relatively new. The Law of Probation, which came into force in July 2012, defines the concept of risk assessment as a standardized investigation aimed at determining both the probability of re-offence and possibilities to reduce it. The current study describes the psychometric variables of the Lithuanian version of the HCR-20 and measures its predictive validity in a correctional sample of violent offenders.
The sample consisted of 74 male offenders from Vilnius, Alytus, and Pravieniškės correctional institutions. The mean age of the population at the initial stage of evaluation was 33.15 years. The number of convictions varied from 1 to 12; convictions for violence varied from 1 to 4. All participants were assessed with the HCR-20, and their data were analyzed in two ways: calculating the total HCR-20 score, and measuring the level of risk based on the weight of risk factors rather than on their quantity. The second stage of the research was conducted after 5 years of approaching the information regarding data on the further convictions of the research participants.
The results have shown that within the period of 5 years, the AUC values for the HCR-20 final risk judgment can significantly predict the risk of both a general as well as a violent crime (respectively, AUC = 0.64 and AUC = 0.73). Therefore, it may be concluded that the HCR-20 is a valid and reliable instrument for assessing the risk of future recidivism (both general and violent). However, it should be noted that a good predictive validity is related to the final risk judgment, while the total arithmetical score is significant for predicting the general recidivism only.