Intimate Partner Violence Risk Assessment Tools: A Review
6. Types of Intimate Partner Violence Risk Assessment ToolsFootnote 5
Risk assessments in cases of intimate partner violence generally fall into one of three models or approaches: unstructured clinical judgment, structured clinical judgment and the actuarial approach. There are also other approaches to assessing risk of intimate partner violence, including victim ratings of risk and the use of risk scales that are designed for assessing the risk of general or violent offending (Hanson et al. 2007). All of the unstructured clinical judgment and actuarial tools have been shown to possess at least some reliability and validity (Kropp 2008).
6.1 Unstructured Clinical Decision Making
The unstructured clinical decision making approach is the most commonly used method of risk assessment (Kropp 2008). In this approach, the professional collects information and renders a risk assessment based on his or her own subjective judgment (Helmus and Bourgon 2011). The professional does not follow any guidelines and the final decisions are justified based on his or her experience and expertise (Guo and Harstall 2008).
One strength of this approach is that it allows the professional to consider the offender’s specific behaviours and circumstances in the development of specific violence prevention strategies (Kropp 2008). It also allows for flexibility (Guo and Harstall 2008). There are, however, a number of limitations to the unstructured clinical decision making approach. Some have argued that this approach lacks accountability, transparency, validity, replicability and relies too heavily on personal discretion (Helmus and Bourgon 2011; Kropp 2008; Nicholls et al. 2006). Others have noted that this approach suffers from low inter-rater reliability and is less accurate than structured risk assessments (Guo and Harstall 2008; Hanson and Morton-Bourgon 2009). Moreover, because this method is heavily dependent on the opinion of the assessor, there is a risk that the assessor may not take into account important information that should be addressed (Kropp 2008).
6.2 Structured Clinical Judgment
The second type of clinical decision making approach is structured
clinical judgment, which has also been referred to as structured professional
judgment and the guided clinical approach (Helmus and Bourgon 2011; Kropp
2008). In the structured clinical judgment approach, assessors follow a set of
guidelines that include specific risk factors that should be considered (both
static and dynamic). These risk factors are determined based on theoretical and
general empirical support (Helmus and Bourgon 2011). The guidelines “
typically include recommendations for information gathering, communicating
opinions, and implementing violence prevention strategies” (Kropp 2008, 207). This
approach does not incorporate the weighting or combining of risk factors, but
rather depends on the evaluator for the final decision of risk (Helmus and
Bourgon 2011; Kropp 2008). The primary goal of the structured clinical judgment
approach is violence prevention, which proponents argue is attained by
including risk factors that are dynamic and can be targeted (Kropp 2008). The
information gathered through this approach can be used to guide treatment
planning and risk management.
The structured clinical judgment approach has a number of strengths.
Supporters have argued that it is more flexible than the actuarial approach (described
below) and because the risk factors are chosen based on empirical evidence, the
tools are more generalizable than tools that were developed based on one
particular population, which is often the case with actuarial tools (Helmus and
Bourgon 2011). In addition, it is more consistent and transparent than
unstructured clinical judgments, but also maintains some of the flexibility and
professional judgment of the unstructured approach (Braff and Sneddon 2007). It
also ensures that a wide variety of information that is specific to a case is
considered and “
allows for a logical, visible, and systematic link between risk
factors and intervention” (Kropp 2008, 207).
There are also a number of limitations to the structured clinical judgment approach. First, because the final evaluation of risk is left to the assessor, the approach is more subjective than basing the decision of risk on a total score (as is the case in the actuarial approach) (Helmus and Bourgon 2011). Moreover, additional empirical research is needed to test the accuracy and reliability of this approach and some studies have found that the inter-rater reliability of the tools is modest (Guo and Harstall 2008; Nicholls et al. 2006). Finally, others have argued that some of the items included in the tools, such as suicidal behaviour, are not strong predictors of intimate partner violence (Hilton and Harris 2005).
There are many intimate partner violence risk assessment tools that have been developed in Canada and the United States based on the structured clinical judgment approach. Three of these tools, the SARA, the Domestic Violence Screening Inventory, and the Danger Assessment, are described in Appendix A.
6.3 Actuarial Approach
The actuarial method is based on the use of predictive, or risk, factors from specific empirical research (Braff and Sneddon 2007; Hilton et al. 2010). These risk factors are assigned a numerical value and a total score is generated through an algorithm (Helmus and Bourgon 2011; Singh 2011). The total score is then used to estimate the probability that the individual will re-offend within a specific time period (Guo and Harstall 2008; Singh 2011). The assessor can also determine the individual’s level of risk relative to similar offenders (Kropp 2008). Training on the proper use of the specific actuarial tools is required (Braff and Sneddon 2007).
Within the actuarial method, there are various approaches used. One
such approach is the pure actuarial prediction approach, as described above.
Another approach is the clinically adjusted actuarial approach, which “
with an established actuarial score and then considers factors external to the
actuarial scheme (i.e., the evaluator is allowed to judge the extent to which the
predicted recidivism rates are a fair evaluation of the offender’s risk)”
(Hanson and Morton-Bourgon 2009, 3). There are currently no known intimate
partner violence risk assessment tools based using the adjusted actuarial
The actuarial approach has a number of strengths. The actuarial approach has been shown to have better predictive ability than the unstructured clinical approach and because actuarial tools use the same criteria, findings can be replicated easily (Hilton et al. 2010; Kropp 2008). Also, the actuarial approach can be used by assessors who do not have clinical training or qualifications, or who are not trained in the area of family violence (Braff and Sneddon 2007).
There are also a number of limitations and criticisms of the actuarial approach. First, Kropp (2008) argues that actuarial approaches do not provide much information about violence prevention strategies and limit the assessor to a fixed set of factors, thereby ignoring case specific factors that should also be considered. Similarly, the tools do not take into account variables that do not occur frequently (e.g., homicidal ideation), but that may be relevant to the case and have been found to be modestly correlated with violence (Kropp 2008; Nicholls et al. 2006). Moreover, assessors may not feel comfortable using the methods as they are limited to using only the actuarial tool when making a risk decision, rather than taking into account other sources of information (Kropp 2008).
The two most commonly used actuarial risk assessment tools for intimate partner violence in Canada, the Ontario Domestic Assault Risk Assessment (ODARA) and the Domestic Violence Risk Appraisal Guide (DVRAG), are described in Appendix A.
6.4 Other Approaches
In addition to the above mentioned types of tools, there are other methods and tools that are used for the purposes of predicting intimate partner violence recidivism. These methods include consulting the victim on their prediction of the offender’s likelihood of recidivism and the use of other risk assessment tools designed to predict general and violent offending.
6.4.1 Consulting the Victim
One approach to assessing risk of intimate partner violence is asking the victim to determine the risk that the offender will re-offend. Research has demonstrated that female victims of intimate partner violence are able to predict their risk of re-victimization with moderate accuracy (Bennett Cattaneo et al. 2007; Hanson et al. 2007; Nicholls et al. 2006). It is recognized that it is important to consult with the victim in these cases, as they are able to provide information pertinent to dynamic factors, such as the offender’s personality, mental health and violent behaviour (Connor-Smith et al. 2011; Kropp 2008). Indeed, the combination of structured risk assessments and victims’ assessments leads to more accurate risk assessments than either approach on its own (Connor-Smith et al. 2011). Depending solely on victims’ assessments, however, is not recommended, as victims may over- or underestimate the risk and may be hesitant to share information for fear for their safety, to protect the offender, or because they do not want to become involved in the criminal justice system (Connor-Smith et al. 2011; Kropp 2008; Nicholls et al. 2006).
6.4.2 Risk Assessment Tools for General and Violent Offending
Another approach to assessing risk in cases of intimate partner violence is the use of risk assessment tools designed to assess general and violent recidivism. The Violence Risk Appraisal Guide (VRAG) and the Level of Service Inventory – Revised (LSI-R) are examples of these general risk assessment tools. Please see Appendix A for a brief description of these tools.
6.5 Factors to Consider When Choosing a Risk Assessment Tool
As can be seen, there are a number of different options available for individuals tasked with conducting intimate partner violence risk assessments. Despite the quantity of intimate partner violence risk assessment tools that are available, there has not been enough empirical research conducted to conclude if one tool, or type of tool, is better than another (Nicholls et al. 2006). Indeed, there has been substantially less research conducted in this area compared to risk assessments for general and violent offending (Helmus and Bourgon 2011).
There are a number of factors that should be considered when determining which risk assessment tool to use. These factors include:
- Whether the goal of the risk assessment is to predict recidivism, to prevent violence, or both.
- Whether the focus is offender-focused or victim-focused. Some tools are designed to assess the offender specifically whether it is to determine the likelihood of re-offending or to develop risk management strategies, while others are focused on whether the victim will be re-victimized.
- Whether the purpose of the assessment is to determine the risk of intimate partner violence or intimate partner homicide (Braff and Sneddon 2007). If the purpose of the assessment is to determine the likelihood of intimate partner violence recidivism, the evaluator could choose such tools as the Spousal Assault Risk Assessment Guide (SARA) or the Domestic Violence Screening Inventory (DVSI). If the purpose is to determine the likelihood of intimate partner homicide, the evaluator could choose the Danger Assessment (DA) or the DV-MOSAIC.
- The time it takes to complete the risk assessment. Some tools consist of fewer items and thus are less time consuming to complete (Braff and Sneddon 2007). For example, assessments that require file reviews, interviews with offenders and other contacts, and reviewing psychological reports, are more time consuming than assessments that do not require these steps. Furthermore, some tools require the input or assessment by a psychologist. One must consider the availability of the psychologist to conduct the assessment and the time it may take him or her to conduct the assessment.
- The individual conducting the assessment (Guo and Harstall 2008). For example, the tool that is chosen may depend upon whether the assessor is a clinician, a law enforcement officer or a social worker.
- The skill and experience of the individual conducting the assessment (Braff and Sneddon 2007). While some research shows that extensive clinical experience is not necessary to complete some measures (e.g., Storey et al. 2011), one should consider the level of clinical experience needed to complete a specific tool when choosing an instrument.
- The setting in which the assessment will take place (e.g., courthouse, women’s shelter) and the accessibility of the information necessary to complete the assessment (Guo and Harstall 2008).
- It is also important to consider what the research literature has found regarding the specific types of assessment tools. In general risk assessment, the unstructured clinical approach has been found to predict risk with less accuracy than structured approaches, often only above chance (Hanson et al. 2007; Hanson and Morton-Bourgon 2009). The actuarial and structured clinical judgment approaches are often compared against each other in the general risk assessment literature; however, as discussed below, research has not been able to determine which type of tool is the most accurate in predicting intimate partner violence specifically. The inability to determine which type of tool is most accurate in predicting intimate partner violence is likely a result of the lack of research that has been conducted in this area (Hanson et al. 2007). Given the lack of research in this area, it is important for assessors to consider the strengths and limitations of the specific types of tools, as well as the other considerations described above.
In addition to the above-mentioned factors, it is important to also consider the general strengths and limitations of intimate partner violence risk assessment tools.
6.6 Strengths and Limitations of Intimate Partner Violence Risk Assessment Tools
There are a number of general strengths and limitations of intimate partner violence risk assessment tools. In terms of strengths, risk assessment tools can provide a common language that can be used among the many professionals involved in cases of intimate partner violence and can provide information to service providers (Braff and Sneddon 2007; Laing 2004). Additionally, intimate partner violence risk assessment tools help ensure that services are allocated to those who need them most and that these services are tailored to the needs and situation of the offender and victim (e.g., safety planning) (Braff and Sneddon 2007; Nicholls et al. 2006). Risk assessment tools also allow for a more accurate method of assessing danger compared to relying solely on professional judgment (Braff and Sneddon 2007). Finally, the documentation involved in conducting risk assessment allows for transparency and accountability on the part of the decision makers and the criminal justice system.
There are, however, several limitations of intimate partner violence
risk assessment tools. First, there are
“no professional standards for (a) the
minimal qualifications of those conducting the assessments; (b) “best
practices” for applying the assessments; (c) training of the assessors; and (d)
evaluation and monitoring of the assessments” (Kropp 2008, 212). Therefore, some
professionals who conduct assessments may be doing so without proper
qualifications, training, or guidance on conducting accurate assessments.
Next, there is some debate amongst experts as to whether tools designed specifically to predict intimate partner violence recidivism are necessary. Many of the risk factors that have been identified as predicting intimate partner violence recidivism are the same as those that predict general or violent recidivism, including young age, unemployment and substance abuse (Hanson et al. 2007; Hanson and Wallace-Capretta 2000; Hilton and Harris 2005). Moreover, Hanson et al. (2007) found in their meta-analysis of intimate partner violence risk assessment tools, that scales designed specifically to assess intimate partner violence recidivism, general and violence risk assessment scales, and victims’ ratings of risk, were similar in terms of their predictive accuracy of intimate partner violence recidivism.
To date, little independent research has examined the reliability, validity and accuracy of intimate partner violence risk assessment tools (Braff and Sneddon 2007; Laing 2004; Roehl et al. 2005). The research that has been conducted has found that these tools have moderate predictive accuracy (Hanson et al. 2007; Kropp 2008). This lack of empirical research has also led to difficulty in determining if one type of tool is more accurate than another (e.g., Guo and Harstall 2008; Hanson et al. 2007). For these reasons, it has been recommended that evaluators consider other sources of information (e.g., the victim’s evaluation of risk) in addition to the use of specific tools (Guo and Harstall 2008).
Finally, risk assessment in general can only allow an evaluator to render a prediction of recidivism and this prediction may be incorrect, which can result in false positives or false negatives (Hoyle 2008; Roehl and Guertin 2000). A false positive that labels an accused as high risk to re-offend when the individual is in fact not, may lead to undue harm to the accused and may heighten the fear experienced by the victim. A false negative that labels an accused as low risk may lead to difficulties for the criminal justice system if the offender does re-offend and may lead the victim to underestimate the risk of re-victimization, which may result in harm to the victimFootnote 6 (Roehl and Guertin 2000).
- Date modified: