It appears that jurisdictions often rely on implementing pre-existing tools derived for similar purposes but on different samples. Given that it is unlikely for a single instrument to have universal applicability, research has suggested that adopted assessments should be piloted and validated on the jurisdiction implementing the tool, since the instrument or its classification scales may not be valid for the agency’s specific population (Gottfredson & Moriarty, 2006; Jones, 1996; Wright, Clear, & Dickson, 1984). Specifically, it should be shown that the instrument can successfully predict the outcomesof interest for the population being served (Flores, Travis & Latessa, 2003; Lowenkamp & Latessa, 2002.)
The literature on pre-trial risk assessment tools developed to date shows that the risk factors measured by these tools generally fall into the following broad domains:
- criminal history
- current legal status
- education
- employment
- housing
- mental health
- relational stability
- residency
- substance abuse
- transportation
Despite these common domains, there is nevertheless a great deal of variety in how the factors that fall within each domain are actually measured from one tool to another. For instance, while Kentucky’s Pre-trial Services Risk Assessment defines stable employment as having held one’s present job for over a year, Monroe County’s (NY) tool defines it as having been employed steadily and full-time for the past three years. Likewise, while all tools include variables related to criminal history, there is variety in which elements of criminal history are included. The Ohio Pre-trial Assessment Tool (2009), for instance, includes age at first arrest. Other tools focus on the number of past convictions. Still others distinguish between felony and misdemeanor convictions.
In addition to the variety within broad domains, there are a number of variables included in some tools that do not appear at all in others. Some examples include whether the defendant expects someone to accompany him/her to arraignment, whether the defendant has a telephone, whether the defendant has citizenship, and whether the defendant is affiliated with a gang. While factors such as these may be predictive of risk in some geographical areas, they are not predictive in all areas.
Furthermore, factors that are predictive of risk at a particular time are not necessarily predictive of risk at a later date. Several jurisdictions, including Harris County (TX), Hennepin County (MN), and the state of Virginia, recently re-validated their pre-trial risk assessment tools and found that the ability of factors to predict risk changes over time. This underscores the importance of not only validating a tool to a particular area but also validating it on a regular basis.
Pre-trial risk assessment tools also vary in how they assign scores to each of the risk variables. Some tools usea small scoring range, such as Ohio’s Pre-trial Assessment Tool, which is scored on a scale of 1-8, while others use a large range, such as Maricopa County’s Release Assessment, which involves a range of 142(+). Furthermore, some tools simply total risk points while other tools utilize both negative and positive points in order to account for protective or mitigating factors. Montgomery County’s Pre-Trial Release Risk Instrument, for example, subtracts a point for defendants aged 50 or older. Tools also vary in how they assign weight to variables. The Virginia Pre-trial Risk Assessment Instrument, for instance, includes eight risk factors, seven of which are worth one point, while the remaining factor—failure to appear, which is highly predictive of risk—is worth two.
Some of the variety seen between pre-trial risk assessment tools is no doubt due to the difference in populations served by the different tools. Some tools are designed to be used in a relatively small geographic area—New York City, for instance, validated a tool of its own. Other tools are designed to be used more broadly—Colorado, in another instance, validated a tool in 10 counties throughout the state. These areas vary not only in size, but also in population density. New York City with its urban composition is quite different from a county in Colorado that comprises both urban and rural areas. In fact, rural areas as a whole are underrepresented in the literature. While they are represented to some degree by studies such as Colorado’s, no studies to date focus exclusively on areas that are primarily rural in nature.
