Timely justice: How case screening improves efficiency, effectiveness, and fairness

Published June 2, 2025

Introduction

This report details findings from the first randomized controlled trial (RCT) regarding intake processes in a local prosecutor’s office in the United States. Evidence from the study shows that promptly reviewing low-level cases and removing those lacking legal sufficiency can significantly reduce the negative impacts of the criminal justice system, provide victims with faster case resolution, and save valuable taxpayer dollars—all without negatively impacting public safety. Over an 18-month period, two part-time screening attorneys cleared 224 cases from the Solicitor's docket, affecting 221 arrestees and saving the state nearly $400,000 in costs associated with jail beds, defense attorneys, and court time.

The experiment stems from Justice Innovation Lab's ongoing partnership with the Ninth Circuit Solicitor's Office (SOL9) and support from the Bureau of Justice Assistance (BJA) and Arnold Ventures. Working with SOL9, Justice Innovation Lab (JIL) previously analyzed racial disproportionalities within the office and authored a follow-up report explaining how a screening process might improve case resolution speed and reduce racial differences in time-to-disposition. This report is the culmination of that effort, providing the results of an 18-month RCT and additional six-month data collection period. The Ninth Circuit's redesign of its internal case processing demonstrates how a data-informed, systems-thinking approach can lead to effective policy making. This is a pivotal and laudable achievement in considered policy change and evaluation.

Why this matters

The criminal justice system affects thousands of lives daily, yet we know surprisingly little about what works in prosecution. Though there is extensive research on what drives crime, and there are many groups that study the impact of prison and sentencing, there is little analytical research with regard to prosecution. This is a massive gap in social policy—while there is an evidential understanding of what drives crime, there is almost no rigorous analysis of society's primary choice of treatment: prosecution. The dearth of research has driven ideological stances in policy making, which in turn has limited innovation in the justice system.

This study arose out of a well-researched and deliberate process to identify inefficiencies and harmful outcomes, develop and test potential solutions, and evaluate results. Such an approach is employed in other policy areas like education and healthcare, but has been severely lacking in the criminal justice space.

What we know (and don't know) about prosecution

Despite the crucial role of prosecutors in the justice system, there is no research comparing the effectiveness of vertical prosecution (where one prosecutor handles a case from start to finish) versus horizontal prosecution (where different prosecutors handle various stages of a case).

However, these serious cases represent just a small fraction of the prosecutor's workload. Most cases in America's criminal justice systems are misdemeanors (minor offenses), and even among felonies, non-violent crimes such as drug possession and property offenses make up the majority of caseloads.

The prosecution process

After police arrest someone, prosecutors must decide whether and how to pursue charging the case based on two main questions:

• Is there enough evidence to prove guilt beyond a reasonable doubt?
• If the evidence is sufficient, do the facts of the case and the person's criminal history support an alternative to prosecution—diversion, specialty court, or a reduced plea offer?

For complex cases, these questions require careful consideration. For simpler cases like shoplifting or drug possession, the legal and factual questions are more straightforward. Yet in vertical prosecution offices, prosecutors often spend considerable time on both simple and complex cases due to overwhelming caseloads—a problem that has worsened since the COVID-19 pandemic.

Why delays hurt everyone

When cases take too long to resolve, the consequences are serious. Though "speedy trial" rules exist to protect arrestees, these protections are often waived, sometimes without people fully understanding the consequences. This leads to extended periods where people may be held in jail before their cases are resolved. Research shows that people held in jail before trial (pre-trial detention) have worse outcomes and are more likely to plead guilty—even possibly when innocent.

Delays can also harm the community as a whole. Recent studies suggest that prosecuting minor offenses may actually increase the likelihood that people will commit future crimes. Research comparing similar defendants found that those who were prosecuted for low-level misdemeanors had higher rates of reoffending than those whose cases were dismissed.

The need for better evidence

While the COVID-19 pandemic forced courts to innovate with virtual hearings and new scheduling practices, and led law enforcement to approach detention in a different manner, these adjustments were made out of necessity during extreme conditions. The extraordinary circumstances make it difficult to evaluate and generalize the impact of these process changes. In addition, there are very few rigorous studies of how changes to prosecution practices and policies affect outcomes. Rather, changes have typically been made without forethought as to how to measure whether the changes are creating the desired impact. Recently though, there have been a small number of researchers conducting high-quality research on various aspects of prosecution, such as:

• Using restorative justice approaches with juvenile offenders
• Removing race identifiers from police reports reviewed by prosecutors
• Testing text message reminders to reduce missed court appearances

Despite this progress, we still lack clear evidence of what works best in prosecution. The findings presented below reveal how utilizing data-informed insights to develop intentional methods for testing the impact of new policies and processes can lead to improved prosecutorial practices and local governance.

Definitions

Below are select definitions of terms used throughout this report. For a further discussion of the screening process and the data used in this report, refer to the "About" and "Experimental design and data" pages. You can also learn more about the prosecutorial process and use of data in prosecution through JIL's criminal justice explainer.

• Randomized controlled trial (RCT): The gold standard for research, randomized controlled trials are used in medical and policy research to measure effectiveness. Generally, an RCT consists of two groups—one that receives the "treatment" and one that does not—with the treatment being the key difference between the groups. In this report, the individuals involved in the RCT were those charged with low-level crimes, like small-amount drug possession or shoplifting below a certain dollar value. These individuals were divided into two distinct groups: the "treatment" group, whose cases were reviewed by a screening attorney, and the "control" group, whose cases followed the usual practice for the office, where the case is immediately assigned to a line prosecutor.
• Line prosecutor: Line prosecutors are government attorneys responsible for prosecuting criminal cases. Line prosecutors have different names in different states—in South Carolina, they are referred to as assistant solicitors, while in other states they are referred to as assistant district attorneys. In this report, we refer to the assistant solicitor assigned to prosecute the case as the line prosecutor.
• Screening attorney: The screening attorney is an assistant solicitor who reviews cases at the earliest possible stage. The screening attorney does not prosecute cases; instead, they may dismiss or remand cases, or refer them to line prosecutors for further evaluation, which can lead to prosecution, diversion, dismissal, or remand.
• Dismissal and remand: Charges can be dismissed, where the charge is dropped and the individual is not convicted. Charges can also be remanded, where the charge is changed to a less severe charge and, in accordance with local laws, sent to a lower court. In this report, we consider cases where all charges are dismissed or remanded to be "non-prosecution" because such cases are removed from prosecution by the Ninth Circuit Solicitor's Office. Though remanded cases may still result in an individual being convicted of a misdemeanor, the punishment is much less severe.
• Diversion: Diversion programs encompass a wide variety of practices and are generally designed to remove cases from a traditional track of prosecution involving jail or prison time. In the Ninth Circuit, diversion refers to a specific set of programs that require evaluation by a diversion coordinator. Because of this practice, screening attorneys did not divert cases, though line prosecutors could divert cases regardless of whether they were screened or not.
• Low-level crimes: The Ninth Circuit Solicitor's Office considers certain drug possession, drug possession with intent to distribute, gun possession, and property crimes to be low-level. These crimes do not involve any violence or physical harm to a victim. Drug crimes are only considered low-level if they involve a small amount of drugs, as determined by the South Carolina legislature.

Findings

This section presents key results from the case screening RCT, highlighting its impact on efficiency, effectiveness, fairness, and savings within the criminal justice system.

Our analysis of all case outcomes is limited to a maximum time frame of six months, since that is the period for which we have complete data for the most recent RCT cases. In our assessment of efficiency, we compare case outcomes over four different intervals: the RCT (May 2023–October 2024), the pilot screening study (March 2022–April 2023), COVID (March 2020–February 2022), and pre-COVID (January 2017–February 2020). Our analyses of effectiveness, fairness, and savings are based specifically on comparisons within the RCT groups.

Efficiency

Total state prosecutorial personnel (prosecutors, investigators, victim advocates, support staff, etc.) in the United States increased by just 4,000 people from 2005 (78,000) to 2020 (82,000), according to the "Prosecutors in State Courts" report series based on the National Survey of Prosecutors. This small, 5% rise, combined with increased demands on prosecutors and the COVID-19 pandemic, has placed significant strains on the system. Adopting policies and practices that alleviate these strains while prioritizing public safety is both efficient and effective.

Improving case processing efficiency benefits everyone involved in a case—victims, judges, prosecutors, defense counsel, and even arrestees. Screening cases serves victims in two main ways. For victims in screened cases, it provides them certainty sooner as to how the case will be handled; for victims of more serious crimes, it gives line prosecutors and other staff more time to devote to their cases. Screening also results in fewer cases and hearings, saving attorney and court time across the board. Justice system actors therefore have more capacity to spend on cases that pose a greater risk to public safety. Low-level cases that are removed from the system get arrestees back on their feet sooner, minimizing long-term harms like job and housing loss and strains on family and social networks.

Results from our study indicate that screened cases (the treatment group) were resolved more quickly than comparable cases that went through the standard prosecution process (the control group). Because there are not clear, agreed-upon standards for case resolution times,

• Local jurisdiction rules require that police submit a police report for an arrest within 30 days. Yet this rule may go unenforced where there is not a screening prosecutor in place to remind officers of the need to submit the report. This attention may explain why screened cases are nearly twice as likely as the control cases to resolve within 3 months (a statistically significant difference of 14.7% compared to 8.8%).
• There is a statistically significant difference in the number of days from arrest to disposition for screened and control cases that are resolved within the first six months. On average, screened cases are resolved within 82 days compared to 106 days for the control cases. With cases resolving nearly a month faster, it is likely that screening saves significant resources for all actors—prosecutors, judges, defense attorneys, arrestees, and victims.

The majority of the time savings in the faster resolution of cases likely comes from a faster time to dismiss or remand cases and, possibly, an overall increase in dismissal and remands for screened cases. Screened cases may be more likely to be dismissed or remanded than control cases for a number of reasons:

• Evidence suggests that pre-trial detention may make arrestees more likely to plead guilty, even when they are innocent. Reviewing cases earlier and removing cases without sufficient evidence alleviates the pressure to plead.
• Screening attorneys may be more detached or objective in reviewing a case and thus more likely to dismiss or remand a case than a prosecutor.
• For the types of cases screened, the screening attorneys have more experience than the line prosecutors receiving the case, and are thus more able to determine when the case is likely to be dismissed later for a variety of reasons.
• In contrast, line prosecutors may develop more information about a case to which the screening attorney would not be privy, and therefore may choose to pursue a case where the screening attorney would not.

Screened cases in the RCT were over 4x as likely to be dismissed or remanded within three months compared to control cases (18.3% compared to 4.2%), as shown in the chart below.

We hypothesized that gains in time-to-disposition would translate into fewer court hearings per case for screened cases. This was based on the assumption that many cases have regularly scheduled hearings that can be "continued"

• We find a small, but statistically significant, difference in the average number of hearings
  for screened cases (2.73) compared to control cases (3.02). Though this difference is small—0.29—over the 1,482 resolved screened cases, this saved 400 hearings in 18 months.

We also expected that the quicker resolution of cases would reduce the case backlog. Overall, however, we do not find a significant decrease in the total case backlog, since the 224 cases removed from the docket by screeners during the RCT account for a relatively small portion of the total case backlog.

As prosecutor caseloads grow in the jurisdiction, improving efficiency is of great importance. The case screening program helps to quickly remove weak cases from the system, allowing prosecutors and the courts to devote more resources to higher priority cases with greater impact to public safety.

Effectiveness

Screening helps to filter out low-level cases and cases with insufficient evidence from the system more quickly. Yet there is a balance to strike with a possible increase in risk to public safety.

To assess the impact of screening on public safety, we compare the rearrest rates for screened and control cases in the RCT, finding that screened arrestees are no more likely to be rearrested than unscreened arrestees. In fact, we find that arrestees with cases dismissed or remanded by the screening attorney have the lowest rearrest rate. This result may be supportive of research finding that prosecuting misdemeanors correlated with an increased likelihood of rearrest. For all calculations below, we only consider individuals that are not in pre-trial detention.

• Overall, 26.4% of screened arrestees were rearrested after their first case in the experiment period compared to 27.2% for the control group. The difference is not statistically significant.
• Given that arrestees have had differing amounts of time to be rearrested, we look specifically at the percent rearrested within six months of the first arrest. We limit to six months since all arrestees have had at least this long to be rearrested after their first arrest in the RCT period. Using this definition, again we find no statistically significant difference in recidivism between screened (15.1%) and control (14.7%).
• To determine the risk from individuals with their case dismissed or remanded by a screener, we also calculate rearrest rates for three distinct groups of arrestees based on their initial case—those dismissed or remanded by the screener, those screened but not dismissed or remanded by the screener, and the control group.
  Again, we limit the recidivism window to six months after the initial arrest. To account for the fact that people are unlikely to be rearrested if in pre-trial detention, we measure the six-month window from the date of release. We find that there is not a statistically significant difference between the groups, though the lowest rate of rearrest is among those with cases dismissed or remanded by the screener.

In examining what individuals were rearrested for, across both screened and control cases, most rearrests were for theft, drug possession, and drug possession with intent to distribute. There were no rearrests for murder or attempted murder. Arrestees in the screened and control groups were rearrested for nearly identical crimes. The majority of rearrests for both groups were for theft, followed by drug possession or possession with intent to sell. Focusing on any differences with respect to crimes of violence:

• 3.5% of rearrests were for assault among the screened group, compared to 3.3% for the control group.
• 4.3% of rearrests were for domestic violence among the screened group, compared to 3.8% for the control group.
• 1.6% of rearrests were for a sex crime among the screened group, compared to 1.5% for the control group.

Overall, we find that screening does not harm public safety. Arrestees in the screened and control groups are rearrested at comparable rates for comparable crimes. In fact, arrestees with their cases dismissed or remanded by the prosecutor are rearrested at the lowest rate, though the difference is not statistically significant. We will continue to monitor this finding as it has implications for evaluating how prosecution impacts public safety.

Fairness

Given the awesome power of the state when arresting someone, and the disruption that an arrest and a conviction has on one's life, it is extremely important that the state exercise its power fairly. By screening cases, the Ninth Circuit addressed an issue that appeared patently unfair—drawing out cases against individuals that were likely to be dismissed simply because overworked prosecutors may not have time to quickly review and dispose certain low-level cases. As outlined above, screening cases provides significant benefits in resolving cases more quickly and limiting system involvement for arrestees. When cases are delayed in resolution, there are likely harms to the arrestee and any victim. As one formerly prosecuted individual reflected—having a pending case for any amount of time places "anxiety that not only is the person going through, but the family of the person." The harm from delays in resolution are greater and starker for people held in jail during their case compared to those who can afford bail and are released.

In South Carolina, after an individual is arrested, they appear before a magistrate judge for a bail hearing. Individuals unable to afford bail and not granted release will continue to be held in the county jail. Given the harmful effects of detention and incarceration, resolving low-level cases earlier in the process limits defendant contact with the justice system, thereby helping arrestees avoid some of the negative collateral consequences associated with detention. In examining differences in jail time, we include all arrestees, even those that do not spend time in jail.

• People who have their cases screened spend an average of 3 fewer days in jail compared to the control group—66 days compared to 69 days—though the difference is not statistically significant. This includes all arrestees who spent time in jail, regardless of the disposition of their case.
• Narrowing the comparison to the time spent in jail for arrestees with their cases dismissed or remanded, the average difference between screened and control cases grows to 10 days—14 days compared to 24 days. This difference is not statistically significant.
• Most importantly, when we compare arrestees with their case dismissed or remanded by the screener the difference is stark. As shown in the figure below, those whose cases are dismissed or remanded by the screener spend 9 days in jail compared to 24 for those in the control group. This difference is statistically significant and represents more than two fewer weeks spent in detention. The significance of this difference to the arrestee cannot be understated, as extended stays in jail increase the likelihood of disrupting an individual's life and might actually drive them to crime.
• Furthermore, arrestees with cases reviewed by the screener, but ultimately dismissed or remanded by the line prosecutor (because the screener did not dismiss or remand) also spend less time in jail than arrestees in the control group. These individuals average 18 days in jail—six days fewer than the control group—though the difference is not statistically significant.

Savings

Using internal data from the Al Cannon Detention Center in Charleston showing the cost per day of detaining an individual being between $100 to $104,

• Using the midpoint of the estimated jail cost per arrestee, the difference in cost per arrestee between screened individuals with their case dismissed or remanded by a line prosecutor compared to the control group is $612 per arrestee. This means that the state saved an estimated $97,000 on the 159 arrestees dismissed or remanded by these prosecutors.
• The savings are even greater when comparing individuals with their cases dismissed or remanded by the screener to the control group. Again, using the midpoint cost difference, the state saved $1,530 per arrestee. Across the 128 individuals affected, the state then saved an estimated $196,000.

Screening half of eligible cases by two, part-time attorneys over an 18-month period then saved Charleston County an estimated $293,000 in jail costs alone.

In addition, we estimate savings from court hearings and attorney salaries using data from the National Center for State Courts

• Taxpayers saved $14,000 in avoided hearings. This is based on the 400 fewer hearings for screened cases at an estimated $35 per hearing.
• Screening attorneys dismissed 140 cases. Arrestees are entitled to a defense attorney, with the court providing an attorney if an arrestee cannot afford one. But for these 140 cases, since the case was dismissed without further court filings, a defense attorney was not required (though one may still have been assigned prior to dismissal. Within the experimental period, 60% of all defense counsel for control and screened cases were public defenders. Assuming that 60% of the 140 dismissed cases then would have been assigned a public defender, that is 84 cases that public defenders did not have to staff. Based on the cost per contract defense attorney of $1,045 per case, the county saves $87,780—equivalent to a year's salary for the median public defender.
• To estimate the savings to the Solicitor's office, we compare the number of cases a part-time screening attorney resolved during the RCT to a group of line prosecutors. For the comparison group, SOL9 provided a list of eight line prosecutors who were likely to handle similar cases during the time period. Over the RCT period, a part-time screening attorney resolved 112 low-level cases whereas a full-time line prosecutor resolved 114 low-level cases. While the full-time line prosecutor has additional, more complicated cases to consider, this is evidence that a devoted screening attorney is able to resolve twice as many low-level cases in a year compared to a full-time line prosecutor.

Based on the above estimates, case screening saved nearly $400,000 in costs associated with jail beds, court time, and defense attorneys over an 18-month period. Additionally, we find that a devoted screening attorney is able to resolve twice as many low-level cases per year as a full-time line prosecutor. The actual savings are likely even greater, since our estimates do not include further savings from bonds, arrestees' ability to retain employment, forensic lab testing, and police officer time attending hearings.

Potential drawbacks

As with all policy changes, there are potential drawbacks that may not have been realized over the time period of the experiment, or that we are unable to measure. We provide those drawbacks below and, to the extent possible, intend to monitor for these moving forward.

• The Ninth Circuit employed two, experienced prosecutors as screening attorneys. One of the attorneys has over 30 years of experience while the other has over 4 years of experience. To our surprise, the more experienced prosecutor dismissed cases more frequently than the less experienced prosecutor. Further research is merited to evaluate the importance of prosecutorial experience in screening to remove the "right" cases.
• If experience does impact the ability of a screening attorney to quickly and properly screen cases, then as screening removes weak cases, non-screening attorneys have fewer and fewer weak cases. As such, newer prosecutors get less experience reviewing weak cases and may be less able to be effective screeners in the future.
• Line prosecutors may become reliant upon screening, or incorrectly believe that screening removes all cases that should not be prosecuted. This may lead line prosecutors to spending valuable time trying to prosecute every case the screener does not dismiss or remand. The screening attorneys are not able to remove all cases that should not be prosecuted since they may not have all information necessary to make a decision at the time of screening.
• There is a disconnect between agencies that save the most from screening and the Solicitor's office, which may cause sustainability challenges. The majority of taxpayer savings are realized by the Sheriff's Office maintaining the jail, the Court in a reduced hearing schedule, and the public defenders office. Yet the Solicitor's Office bears the entire cost of employing the screening attorney, and given the long-term underfunding of prosecutor offices, this may create budgetary challenges moving forward.

Conclusion

The Ninth Circuit Solicitor's Office took a careful, considered approach to changing its case intake process to address a need. After 18 months of measurement and 6 additional months of data collection, this approach created clear and convincing evidence that screening low-level cases saves taxpayer dollars without harming public safety. Across a variety of metrics, screening demonstrates that prosecutors can quickly remove low-risk cases from the felony docket in a manner that greatly reduces negative impacts on arrestees while giving staff more time to focus on serious cases with victims. Furthermore, screening easily pays for itself—two part-time attorneys generated enough cost savings in the system to cover their salaries twice over.

Other jurisdictions face similar challenges to the Ninth Circuit and can benefit from replicating the screening program, as well as the approach to making changes based on data-informed decision-making and a well-thought-out process for evaluation. While many large jurisdictions may already have a dedicated screening unit, smaller jurisdictions may not have the resources to do so or may not have considered screening as a way to address case backlogs. This report builds evidence for prosecutorial best practices and provides guidance on how an office can innovate to improve the justice system for all involved while addressing public safety challenges in a cost-effective manner.

Appendix

As a robustness check to our rearrest calculations, we calculate monthly calls for service and arrests in Charleston County from 2022 to 2024 based on publicly available data. As shown in the graph below, though the average number of monthly calls increased in 2024 compared to 2022 and 2023, the difference is not statistically different for either year.

In the case of arrests, there is a statistically significant increase in the monthly number of arrests starting for 2024 compared to 2022 and 2023. As described above, the increase in arrests is not of individuals that passed through screening and is not indicative of a rise in violent or more serious property crime in Charleston.

In examining what crimes are driving the increase in arrests, we find that they are non-violent offenses, many of which are not prosecuted by the Ninth Circuit. These arrests are not connected to screening, as the rearrest rates described above do not indicate an increase in crime whereby screened individuals are being rearrested at significant rates. Below are the top five charges with the greatest increase in arrests from 2023 to 2024. The increase in arrests for these minor offenses may represent a change in enforcement strategy by police, but does not suggest a rise in serious crime.