Dealing fairly with pretrial detainees

Using risk-based assessments, we could improve justice, lower crime, and save taxpayers money

IT SOUNDS TOO GOOD TO BE TRUE, but there is a tested way to improve the fairness of our justice system while saving taxpayers money and lowering crime. The solution is to make data-driven decisions about individuals who get arrested to determine which ones should remain in jail while awaiting trial and which ones should be released.

Massachusetts currently has no standard way to assess the risk a defendant poses while awaiting trial – either the risk of not returning for the scheduled court date, or the risk of committing a crime while out on bail. Instead of being based on risk, the system is based on the affordability of bail. As a result, many low-risk pretrial defendants are spending unnecessary time in jail, typically because they are too poor to afford their bail amount, even when it’s as low as $300. Higher-risk defendants, meanwhile, get released even when they are a danger to society, simply because they can afford their bail.

The lack of data-driven decision-making in the pretrial process is a national problem. Across the United States, 90 percent of jurisdictions operate without a risk-based bail system, leaving judges to set bail using their gut instincts or by following a schedule setting bail amounts for specific crimes. The ramifications are enormous. It costs an estimated $14 billion every year to jail the 7.4 million pretrial detainees who have been arrested and are awaiting their day in court. Pretrial detainees account for 95 percent of the growth in the jail population over the last 15 years, and represent 62 percent of all inmates held in local jails.

Kalief Browder was a victim of this system. Accused at age 16 of stealing a backpack but insisting on his innocence, Browder could not pay his $3,000 bail and was jailed at Rikers Island in New York City. He stayed there for 1,111 days, from the time he was 16 until he was 19, without his case ever coming to trial. The charges against him were eventually dropped, but he lost three years of his youth and he ended up taking his own life this past summer at the age of 22.  Taxpayers funded the $510,545 cost of his pretrial incarceration.


Dangerous repeat offenders often get out when decisions are based on dollars instead of data. In November 2009, four police officers in Lakewood, Washington, were sitting at a café, working at their laptops when they were killed, execution-style, by a man who had been released from jail not once but twice while awaiting trial on rape charges involving a 12-year-old girl. During his short stay in jail, he threatened to kill jail guards, and a psychological evaluation deemed him a threat to the community.  But he paid his bail and was allowed back on the streets again.

One of the national leaders in using data to make pretrial detainment decisions is Washington, DC. All those arrested in DC are assessed for their risk and that’s what determines whether they are detained in jail or released while awaiting trial.  The risk assessment tool gathers 38 data points, which are then verified by interviewing the defendant.  Judges, who have access to this rich data profile before meeting the defendant, can more effectively balance community safety and the individual’s liberty in making detention decisions.

Some of the types of data included in the risk assessment include the defendant’s community ties, criminal history, prior court appearance rate, citizenship status, physical and mental health status, substance abuse status, and any existing requirements for probation or parole.  Each element in the risk analysis is based on research and scientific evidence. The best risk assessment tools are locally-validated, and the DC one is.  In fact, the DC tools have been updated since they were first created to reflect improvement in the precision of prediction of risk.

Based on the quantitative risk score, a defendant is classified as either high risk, medium risk, or low risk. For high-risk defendants, jail is typically the best option.  For low-risk defendants, many can be released on their own recognizance. Medium-risk defendants can be supervised in the community in a number of ways to improve the likelihood they will return to court and remain crime-free in the interim. Pretrial supervision can include living in a half-way house and/or wearing an electronic monitoring device.

The results are impressive — 85 percent of defendants are released before trial, either on their own recognizance or with supervision. Of those, 11 percent were rearrested while out on bail and 11 percent failed to appear for their court date; the national re-arrest rate for those released before trial is 29 percent, while 17 percent fail to appear in court.
Other states and cities using risk-based tools to set bail have had similar results. Kentucky now releases 70 percent of defendants pretrial, yet defendants are twice as likely to return to court and remain arrest-free as those in other jurisdictions. In New Orleans, a study found that 95 percent of those released under a risk-based bail system show up for court and 96 percent remain arrest-free while awaiting their trial date.

Maine estimates it saves $2 million a year using its risk-based bail system. New Jersey, where at one point 73 percent of those in jail were awaiting trial, is moving to a risk-based bail system next year. A study there indicates “the greatest opportunities to responsibly reduce New Jersey’s jail population are related to more efficiently and effectively managing the pretrial population.”

Massachusetts lags far behind these other states. It has no accurate data on the number of pretrial detainees and the cost to hold them. There is also next to no information on how many detainees released on bail return for their court dates or are arrested while out on bail. In short, there’s not enough data to hold a fact-based debate about what we are doing now and whether the outcomes are acceptable.

A report issued last year in Middlesex County was a step in the right direction. It said the average daily pretrial population at the Middlesex County Jail increased 30 percent between 2008 and 2013, rising from 593 to 772. The increase is not surprising, as pretrial detention rates are increasing across the country. The Middlesex report indicated 25 percent of pretrial detainees were being held on drug and property charges and 23 percent were being held for $1,000 cash bail or less.

“The nature of these charges and bail amounts suggest that these defendants are ‘release-able’ in a general sense,” said the report, which called for broader use of  “validated risk assessment instruments” capable of distinguishing between “low and high-risk persons in terms of pretrial risk, domestic violence risk, and recidivism risk.”

The Middlesex County report was amended in April after some of the data in it was criticized. The amended report lamented the poor quality of data available on jail operations but said the conclusions reached in the original report remained valid, particularly the call for risk-based, bail-release decisions.

A good guess of how much money the state could save by moving to a risk-based bail system is $60 million a year. It’s only a guess because the data are so poor, but I came up with it by assuming that 60 percent of the state’s jail inmates are being held pretrial at a cost of roughly $300 million a year. Based on other jurisdictions reducing their jail population by 20 to 67 percent with pretail reform, I conservatively estimate Massachusetts could reduce its pretrial population by 20 percent with a risk-based bail system, leading to savings of $60 million.

Any cost savings would be achieved only over time because of the high fixed costs in corrections (facilities, collective bargaining agreements, etc.).  There would be additional costs to supervise released pretrial detainees, but those costs would be significantly lower than the cost to incarcerate.  Nationally, the cost of supervising someone on pretrial release is 90 percent lower than the cost to incarcerate them ($7.17 per day for supervision versus $74.61 to detain).

In addition to the cost-savings benefits, a data-driven approach to pretrial decisions would have other intangible benefits, including:

Fairness Using data from a risk-assessment tool means that there is fairness across the system from city to city and from courtroom to courtroom. This levels out subtle biases that may affect the fairness of the administration of justice.  Research shows that defendants who already have advantages (higher income, employment) are released more often than less-advantaged (lower income, minority) peers, for the same crime.

Transparency  By using data instead of opinion to guide decision-making, and by having data to document decisions, public trust in the system may increase.

Humanity Pretrial release of low-risk defendants would allow people who pose little risk to the community (and who are supposed to be considered innocent until proven guilty), to show up at work, have dinner with their families, and stay on track with their responsibilities.

Meet the Author

Jane Wiseman

Founder, Institute for Excellence in Government
The origins of the statute governing bail and the pretrial process in Massachusetts date to 1836, when Andrew Jackson was president, the railroad was new, slavery was legal, and the Boston Police Department did not exist. While a hodgepodge of updates have been made over the years, the law is ready for a comprehensive overhaul.  An ambitious bill crafted by Rep. Tom Sannicandro of Ashland and Sen. Ken Donnelly of Arlington aims to bring Massachusetts into the 21st century, leveraging best practice research on what works. The bill would establish a long-overdue data-driven approach to the pretrial process. Its passage would represent a long overdue change from the status quo.

Jane Wiseman is the founder of the Institute for Excellence in Government, a nonprofit consulting firm based in Boston devoted to improving government. She worked previously as the assistant secretary of public safety in Massachusetts and special assistant at the National Institute of Justice, US Justice Department.