For nearly five decades, CJA has produced rigorous, cutting-edge research on criminal justice issues in New York City. CJA’s publications have examined topics such as bail and pretrial release, juvenile justice, court date notification, and desk appearance tickets.
In this continuing series of “From the Archives,” we revisit some of CJA's classic research studies and highlight how their findings are relevant today. Each study is published in full on CJA’s website and is accompanied by new analysis that contextualizes the study's results alongside new knowledge and contemporary developments. This post of “From the Archives” reviews the findings of Mary Phillips Arraignment Observation Project as presented in a series of 2004 reports¹ with comparisons to recent findings of CJA’s current Court Observation Project with data from October 2022 through March 2023.
Introduction
Most studies of bail and release decisions at first court appearance have relied on statistical analysis of official court data including, for example, the severity of the charges, details of any criminal history and, perhaps, some characteristics of the persons facing criminal charges. In contrast, Mary Phillips ground-breaking work examined the negotiations between the decision-makers at arraignments in Criminal Court through in-person courtroom observations. The Arraignment Observation Project began as a pilot program in 2001 and 2002 with CJA researchers observing arraignments in Manhattan courtrooms. The pilot program demonstrated that the observations could yield useful data. Some of the observation data showed considerable potential in explaining why charged persons with similar criminal records who were facing similar charges did not always receive the same outcomes in terms of release and bail setting. The pilot was so promising that the project was extended to include Brooklyn as well as Manhattan and the researchers were frequently granted permission to sit next to the judge at the bench instead of in the audience, which resulted in enhanced audibility of the proceedings.
The Court Observation Project, which began in 2017, expanded the initial project by employing CJA’s trained Pretrial Associates (PTAs) to observe arraignments citywide. Since the current Court Observation Project commenced, CJA implemented a revised release assessment in 2019², and changes were mandated by NY State’s 2020 bail reform law and subsequent amendments³. In light of these changes, we focus on recent findings, comparing observation data collected from October 2022 through March 2023 to findings in the 2004 report.
About Arraignment
An arraignment is an adult’s first appearance in Criminal Court following an arrest. Many cases are disposed at that hearing, either by plea, dismissal, or other resolutions. The remaining cases are adjourned (also referred to as continued), which means that a next court date is set for a follow-up hearing. The arraigning judge must decide whether to release the person with no conditions (release on recognizance, referred to as ROR), set some conditions such as supervised release, remand the person to pretrial detention with no bail set, or to choose an amount of money bail that the person must post to secure pretrial release. The judge makes the pretrial release decision with input from the prosecutor (also referred to as the assistant district attorney or ADA), the defense counsel, and the CJA release recommendation.
Arraignment Observation Study Findings 2004
The central finding of the 2004 study based on roughly 2,000 observed arraignments in Manhattan and Brooklyn between September 2002 and March 2003 is that the prosecutor dominated release and bail decisions:
- The prosecutor bail request was the most powerful predictor of both ROR and, if ROR was not granted, of the amount of bail set.
- The CJA release assessment (RA) recommendation had a small but significant effect on the ROR decision even though it had no effect on the prosecutor request nor on the amount of bail set.
- The severity of the charge persons faced (“charge severity”) primarily affected prosecutor requests for bail.
- In addition to the severity of the charge, whether the person had previously failed to appear at a scheduled court date was an important determinant of the prosecutor consent to ROR.
- When the prosecutor request conflicted with the CJA release recommendation, the judge was more likely to side with the prosecutor, especially in the uncommon situation when the prosecutor favored ROR but the CJA release assessment did not.
October 2022 to March 2023 Court Observation: Changes Since 2002 to 2003
A lot has changed in the justice landscape in the twenty years since the data was collected for the 2004 Arraignment Observation Study.
First, as mentioned above, the release assessment (RA) recommendation used by CJA was revised in the fall of 2019. Both recommendation systems are based on empirical research identifying objectively measured factors that are associated with failure to appear (FTA). In other words, both employ a validated point scale system to evaluate the likelihood that individuals, if released, will appear for subsequent court dates. The largest difference between the RA in effect in 2002 to 2003 and the current RA, at least for purposes of comparing arraignment court observation outcomes, is that the current release assessment (RA) is scored more leniently when the arraignment charge is a misdemeanor or non-violent felony and is more restrictive when the charge is a violent felony offense (VFO). Notably, the updated RA can recommend “Consider all options” which can include an outcome of supervised release, or release under supervision (RUS). Both the release assessment recommendation of “Consider all options” and an RUS outcome were not available at the time of the original Arraignment Observation Study.
Second, NY State’s bail reform law and its amendments⁴ Bail Reform Amendments.” May 2023. eliminated the use of cash bail for nearly all misdemeanor cases and most cases with non-violent felony charges, thus enabling most persons to be released with no restrictions, under supervision, or with electronic monitoring. In 2002 to 2003, persons facing criminal charges in NY State could be detained pretrial or released on recognizance (ROR) or on bail. In both time periods, the amount of bail when set is at the discretion of the judge.
October 2022 to March 2023 Findings and Comparison to 2002 to 2003
CJA Pretrial Associates (PTAs) observed 16,602 arraignments for Summary arrests (those held in custody from arrest to arraignment) that were continued (not resolved) at arraignment across the boroughs in the six-month research period⁵. Since the 2004 study was conducted in Brooklyn and Manhattan, only the 8,134 observations in those boroughs are included in the comparisons presented below.
Also, considering changes in the CJA release assessment recommendation and those due to implementation of NYS bail reform, the findings are presented separately for cases with violent felony charges (VFO), other felony charges and misdemeanor (or lesser) charges. Thus, the findings for 2002 to 2003 are a re-analysis of the 2004 findings.
CJA Release Recommendation
The CJA release assessment recommended ROR in just over eight of every ten cases observed in Brooklyn and nearly seven of every ten in Manhattan in 2022 to 2023, much higher than the 66% in Brooklyn and 53% in Manhattan recommended for ROR under the assessment system in place during the earlier project (Figure 1). The proportion of cases recommended for ROR was higher in Brooklyn than in Manhattan for each charge severity category in both time periods and was also higher in 2022 to 2023 than in the previous period for each charge severity category.
As previously mentioned, in 2022 to 2023 the CJA release assessment can recommend “Consider all options” when the client does not qualify for a recommendation for ROR. “Consider all options” was recommended for 6.5% in Brooklyn and 8.5% in Manhattan (data not shown).
Prosecutor Release Request (ROR or RUS) When CJA Recommended ROR: 2022 to 2023
After the implementation of bail reform, misdemeanor-level cases and most non-violent felony cases were not eligible for bail. Now the only release status options available in most of these cases is ROR or release under supervision (RUS). As mentioned, RUS was not an option at the time of the original Arraignment Observations Project.
In 2022 to 2023, the prosecutor recommended release under supervision (RUS) when CJA recommended ROR in 15% of observed arraignments in Brooklyn and in 27% in Manhattan. It is noteworthy that requests for RUS are higher in Manhattan than in Brooklyn in each charge severity category, reaching a high of 46% for non-violent felony cases, but also for 23% of misdemeanor-level cases which are eligible for ROR. Figure 2 displays the prosecutor release requests when CJA recommended ROR and shows that there is far more agreement between the CJA recommendation and the prosecutor release request in Brooklyn than in Manhattan.
Prosecutor Agreement When CJA Recommended ROR
While the increase in the portion of cases recommended for ROR at arraignment by CJA is striking, the increase in the portion of cases in which the prosecutor also requested ROR is even more dramatic: The prosecutor agreed with the CJA ROR recommendation in 56% of observed arraignments in Brooklyn in the recent project compared to only 23% in the 2002 to 2003 study and 48% agreed in Manhattan compared to only 27% in observed arraignments in the earlier period (Figure 3).
As expected, in both time periods and both boroughs, prosecutors agreed with the CJA ROR recommendation more frequently when the case was a misdemeanor than when the charges were more severe. Overall, agreement rates for cases with misdemeanor charges were higher in the recent study than in 2002 to 2003. Agreement rates for misdemeanor charges were higher in Brooklyn (77%) than Manhattan (62%) in the recent study and higher in Manhattan (42%) than Brooklyn (31%) in the earlier study. When the arraignment charge was a violent felony (VFO), prosecutors agreed with the CJA ROR recommendation in 21% of cases in Brooklyn and 13% in Manhattan in 2022 to 2023. Conversely for the same type of cases (VFO) in the earlier study, prosecutors agreed with the CJA ROR recommendation only 4% in Brooklyn and 17% in Manhattan.
Prosecutor Agreement/Disagreement with the CJA Recommendation
Figure 4, below, displays the prosecutor (ADA) agreement/disagreement with the CJA release assessment (RA) ROR recommendation for each borough across both time periods for the three charge severity categories. For purposes of this comparison, we treated CJA and prosecutor requests for RUS as requests for ‘No ROR’ (i.e. release with no conditions). We have already seen the rate of agreement among the prosecutor and CJA on an ROR recommendation: 56% in Brooklyn and 48% in Manhattan in the current project compared to the agreement rate of 23% and 27% in Brooklyn and Manhattan, respectively, in the 2004 findings. As shown in Figure 4, CJA and the prosecutor also agreed on ‘No ROR’ (light blue) in 14% of observed arraignments in Brooklyn and 26% in Manhattan in 2022 to 2023, far lower than in the 2002 to 2003 data (29% in Brooklyn and 39% in Manhattan), even though the RUS requests are included here as ‘No ROR’. In addition, observed arraignments in which the prosecutor and CJA agreed on a No ROR recommendation (‘ADA agrees with CJA (RA – No ROR)’) were far more frequent in Manhattan than in Brooklyn in both time periods for all three charge severity categories. This rate of agreement of a No ROR recommendation was especially high in the earlier period in Manhattan cases with non-VFO charges where CJA and the prosecutor agreed in nearly half of the cases (41%).
CJA and the prosecutor often disagreed. The rate of disagreement between the prosecutor and CJA when the CJA release assessment recommended ROR (‘ADA Disagrees with CJA (RA – ROR)’; light orange) was 36% in both boroughs in 2022 to 2023 compared to 51% in Brooklyn but 39% in Manhattan in 2002 to 2003. Higher rates of prosecutor disagreement when CJA recommended ROR in the earlier study are apparent for each charge severity level but are especially marked in Brooklyn VFO cases: the ADA disagreed with the CJA ROR recommendation in 71% of VFO cases in 2002 to 2003 and 64% in 2022 to 2023.
The prosecutor rarely requested ROR when CJA did not recommend ROR (orange). Only 5% to 8% of arraignments in both boroughs in across the two studies had this outcome, save for the 14% of observed arraignments for misdemeanors in Manhattan during the earlier study.
The Judicial ROR Decision
Figure 5 displays the judicial ROR decision by prosecutor and CJA agreement or disagreement in each borough. In both projects and in both boroughs, the judge nearly always decided to ROR when both the prosecutor (ADA) and CJA agreed to ROR (blue). Although it occurred rarely, the judge sometimes ROR’d at arraignment when both the prosecutor and CJA agreed against ROR. In 2022 to 2023, CJA and the prosecutor did not support ROR (‘ADA Agrees with CJA (RA – No ROR)’; light blue) between 1% and 5% of all observed arraignments where the judge ROR’d across all charge severities in both boroughs. Judges were more likely to override the lack of support for ROR from both the ADA and CJA in the earlier project (16% in Brooklyn and 26% in Manhattan).
The judge frequently decided on ROR when the prosecutor request and the CJA recommendation differed. In both projects and in both boroughs CJA recommended ROR and the judge ROR’d (light orange) more often than when the prosecutor recommended ROR and the judge ROR’d (orange). It is important to note, however, that the ADA recommended ROR despite CJA not recommending ROR (i.e. recommending “Consider All Options” or RUS) in very few observed cases.
As mentioned earlier, release under supervision (RUS) was not available during the earlier court observation project but is a frequent release decision since the implementation of bail reform. Figure 6 displays the judicial decision to ROR or RUS in the 2022 to 2023 project when the ADA and CJA disagreed about ROR. As shown below, judges often decided to RUS at arraignments when the ADA did not support ROR despite a recommendation for ROR from CJA (63% in Brooklyn and 52% in Manhattan across all charge severities). In both boroughs, the prosecutor disagreed with the CJA ROR recommendation at a greater rate in arraignments that resulted in a judicial RUS compared to arraignments where the judge decided on ROR.
Figure 7 shows the failure to appear rate (FTA) when the judge ROR’d at the observed arraignment by CJA/prosecutor agreement or disagreement in their release recommendations. FTA is measured as failure to appear as scheduled at the first scheduled court date after arraignment. However, since the implementation of bail reform, nearly every instance in which a person misses the first court date results in a ‘stayed’ (not issued) bench warrant. To account for this change, the calculation of FTA rate must also reflect attendance at the second scheduled court date after arraignment if there was a stayed warrant at the first.
The findings show very low rates of FTA after the judicial ROR decision when both the prosecutor and CJA agreed on ROR, about 5% across the charge severities. In the small number of cases in which the judge decided to ROR despite prosecutor/CJA agreement against ROR, the FTA rate was 34%, ranging from 22% to 35% across charge severities.
The most important findings concern FTA rates when the judge decided to ROR when CJA and the prosecutor disagreed. FTA rates were quite high when the judge ROR’d in agreement with the prosecutor but the CJA release assessment indicated that the person did not score high enough to qualify for a recommendation for ROR 21% overall, 12% for VFO, 10% for other felonies, 23% for misdemeanors). In contrast, FTA rates were quite low when the judge ROR’d in agreement with the CJA release assessment, despite the prosecutor’s advocacy for more restrictive release conditions. Only 6% of those released at arraignment did not appear at the first opportunity when CJA recommended ROR and the prosecutor did not: 1% in VFO cases, 5% in other felony cases, and 7% in cases with misdemeanor or lesser charges.
The very low rate of FTA when the judge decided to ROR in accordance with the CJA release assessment despite prosecutor disagreement could only be revealed by the work of the CJA pretrial associates (PTAs) observing arraignments and carefully recording their observations.
Future Research and Implications:
Future research should follow-up in several directions.
First, the new research should follow the lead of the 2004 project which produced multivariate statistical models that accounted for the relative effects of different factors on the release decision. Using data reflected in the CJA release assessment as well as data that CJA routinely receives from NYPD (such as charge severity, client age, race, and gender), future research should assess the effects of potentially confounding factors.
Second, multivariate modelling should include case eligibility for bail (i.e. whether the top charge in a case qualifies for bail). This would allow for the rates of agreement and disagreement between the prosecutor request and the CJA release assessment recommendation to control for the effects of bail reform and its amendments. Further, the specifics of the prosecutor request and the defense counsel request should be explored and potentially included in the models.
Additionally, the court observation project data is extremely rich. It includes the specifics of the prosecutor bail requests, including the amount and form of bail, as well as reasons presented in court in support of release requests. New court observation research could expand to include more detailed information regarding prosecutor bail requests, such as what areas pertaining to the client are discussed in court by which parties, and for what purpose (e.g. mention of behavioral health needs, housing, family present at court, etc.).
Finally, new research should explore the court appearance record of persons who were released under supervision. An analysis of FTA rates for individuals who the judge released with or without conditions by all of the combinations of CJA/ADA release requests would provide unique insight into the release decision at arraignment in light of bail reform.