The Dangers of Considering “Dangerousness”: The History & Implications of Risk Assessment-Based Pretrial Decision-Making
By Kate Jassin, Research Associate, and Cecilia Low-Weiner, Research Associate
This blog is part of a series on the implementation, effects, and challenges of bail reform in New York State. Check out “Bail Reform in New York” for an overview of bail & other criminal legal reform legislation in the state as well as other resources & materials.
Legal decisions that involve the potential loss of a person’s liberty requires careful balancing of both community safety and individual due process rights. Founded on a presumption of innocence, the criminal legal system was meant to rely on detention only as a last resort, yet the inability to pay bail is one of the main drivers of incarceration nationwide: up to two-thirds of individuals held in local jails have not yet been convicted of a crime.
In 2020, legislation took effect in New York State that sought to reduce unnecessary pretrial detention by eliminating bail for most misdemeanors and non-violent felonies, as well as requiring judges to use the least restrictive conditions to ensure individuals return to court. Those critical of the legislation—including some media outlets, law enforcement officials, and politicians—have linked recent upticks in violent crime to bail reform, despite limited evidence that the legislation is a meaningful cause as compared to pandemic-related impacts on the community. This has led to a push, including by New York City Mayor Eric Adams, to reverse the state’s historical precedent and mandate that judges consider “dangerousness” — an assessment of a person’s likelihood of committing a new offense — in deciding whether to hold someone in jail during their trial.
These considerations are of great consequence. Research has linked pretrial detention to numerous negative consequences, including increased pressure to plead guilty, greater chance of conviction, harsher sentences, and loss of employment, housing, and parental rights. It may even increase chances of legal system involvement in the future. Due to existing racial and ethnic disparities in the legal system, these consequences would disproportionality impact low-income and communities of color. While “dangerousness” assessments are touted as an effort to reduce reliance on pretrial detention by detaining only those deemed a risk to public safety, very little is known about whether these considerations in bail decision-making would reliably decrease pretrial detention or reduce racial and ethnic disparities, the very two things the original legislation was meant to address.
Research has linked pretrial detention to numerous negative consequences, including increased pressure to plead guilty, greater chance of conviction, harsher sentences, and loss of employment, housing, and parental rights. It may even increase chances of legal system involvement in the future.
While the most recent version of the bail legislation proposed in the budget stopped short of Mayor Adams’ full proposal to include dangerousness in bail-setting decisions, it nonetheless mandates judges to consider public safety-based bail considerations, such as gun possession/use and whether an individual is charged with a crime that caused “serious harm.” Given this shift towards public safety-based considerations and its potential for more preventative detention, it is critical that stakeholders understand the full history and implications of “dangerousness” in pretrial release decisions, and its impacts on public safety, incarceration rates, and racial disparities.
Defining and Predicting Dangerousness
Today, New York’s bail statute dictates that judges can only set bail to ensure an individual returns to court. They cannot take into consideration a person’s “dangerousness,” or the potential harm they may pose to the community. This has been the standard since 1970, when the State Legislature passed bail reform laws that explicitly excluded a “dangerousness” clause—one of the only states at the time to do so. This statute has remained largely unchanged until 2019, when the new bail reform legislation was signed into law. This new legislation once again omitted dangerousness considerations, instead focusing on limiting judicial discretion, considering ability to pay, and ensuring the “least restrictive” means were imposed to ensure individuals return to court.
Defining dangerousness, like predicting it, has been a subject of debate. Approximately half of the states that explicitly reference dangerousness use a vague definition, including phrases like “safety of the community” and “danger to the public.” One of the main ways that dangerousness is functionally measured in bail determinations is by risk assessment tools that seek to determine someone’s risk of flight and the likelihood they will be rearrested before their case is resolved. These tools sort people into “risk” categories, from low to high, either by using clinical interviews or—now the more common approach—an actuarial algorithm. Some view these as an important tool for facilitating pretrial release while maintaining public safety, while others argue that they perpetuate racial bias and counterproductively increase pretrial detention.
Algorithmic risk assessment tools attempt to predict outcomes based on historical data such as occupational status, education background, drug or alcohol use, and prior criminal record. However, the accuracy of these assessments is not well-evidenced. Studies have indicated that it remains very difficult to accurately predict whether individuals will commit crimes in the future. Other studies have shown that algorithmic tools are no better at predicting risk than people provided with short descriptions of those arrested. And while the tools themselves may be touted as “race-neutral,” racial bias may be baked into the algorithm due to disparities in the criminal legal system outcomes used to assess risk, like rates of arrest and conviction. Additionally, these algorithms are trained on data prior to bail reforms and therefore don’t take into account the risk-reducing benefits of these reforms.
And while the tools themselves may be touted as “race-neutral,” racial bias may be baked into the algorithm due to disparities in the criminal legal system outcomes used to assess risk, like rates of arrest and conviction.
Proponents of including a dangerousness assessment argue that these risk tools can assist judges and other criminal legal stakeholders in more precisely identifying the small percentage of individuals who are likely to be rearrested for violent crimes after release. However, there is limited evidence that such risk assessment tools reliably predict future violent crime. Multiple studies have found that these instruments systematically overestimate risk, with the vast majority of people facing trial classified as high risk having not been rearrested after release. This means increased reliance on these tools would likely expose more individuals to preventive detention than necessary—which, as stated before, can have lasting impacts on them, their families, and the community.
Implications of Considering Dangerousness
Given the increasing attention “dangerousness” assessments are receiving, it is important to consider the implications of using such tools. Dangerousness considerations would, in effect, promote a form of preventive detention that occurs prior to conviction, tipping the balance in favor of public safety concerns over individual liberty and due process. In times of rising violence, it is understandable why policymakers would be compelled to consider such a choice. However, this can be seen as a violation of the presumption of innocence, and evidence that pretrial risk assessment tools are not accurate predictors of future crime would imply the trade-off may not be worth it.
When considering the inclusion of dangerousness assessments in pretrial release decisions, it is important to review the evidence of its effectiveness in reducing crime, the impacts on pretrial detention rates, and its effects on racial disparities in both pretrial detention and case outcomes. Notably, available data suggests that pretrial release decisions under the new bail reforms are not a primary driver of the recent uptick in violent crime in New York. Similarly, New Jersey, which had recently eliminated cash bail, significantly reduced their jail population with no meaningful increase in crime rates. However, unlike New York, New Jersey replaced cash bail with an approach that included an algorithmic risk assessment tool for pretrial release decision making. While the state did experience a significant decline in its jail population, they were unsuccessful in reducing racial and ethnic disparities. Kentucky also mandated the use of pretrial risk assessments in release decisions, and did not see any substantial impacts on releases, court appearance rates, pretrial crime, or racial disparities in pretrial detention. Such mixed results are likely due to wide variations in how risk assessment tools are validated and used by judges in each jurisdiction, suggesting that the accuracy and effectiveness of pretrial risk assessments is much more difficult to predict across contexts.
Notably, available data suggests that pretrial release decisions under the new bail reforms are not a primary driver of the recent uptick in violent crime in New York.
Perhaps the most consistent findings regarding pretrial risk assessments is that the implications of these inaccuracies will have a far greater impact on Black people facing trial, who are disproportionately misclassified as high risk. This suggests that any increases in rates of pretrial detention that may result from the use of risk assessment tools would be disproportionately higher for Black individuals, exacerbating racial disparities.
Next Steps: Further Study
Ultimately, there isn’t a simple solution when it comes to assessing risks to public safety. As the evidence suggests, pretrial release decisions that account for “dangerousness” may undermine efforts to reduce unnecessary detention, increase racial disparities, and not address public safety in the ways proponents suggest. Instead, policymakers and the public should exercise caution in diagnosing the cause of the rise in violent crime, and allow for sufficient time to study the impact of existing bail laws before making changes that may heighten already existing issues of over-incarceration and racial disparity.
Photo by PaeGAG on Adobe Stock.