OBJECTIVE We examined efforts by a Mississippi court to base pretrial release decisions on risk assessment rather than primarily on bond. HYPOTHESES (a) Pretrial detention will be shorter than that associated with prevailing bond practices in the same counties. (b) Rearrest rates will be lower than a similar pretrial population in a nearby southern state. (c) False positive rates for predicting rearrests will be higher for African American than Caucasian participants. (d) Pretrial detention will be longer for African American participants because of higher risk scores or assessment overrides. METHOD Pretrial defendants (N = 521) completed the Risk and Needs Triage (RANT) within 2 weeks of arrest, and outcomes examined included the length of pretrial detention, index case dispositions, and rearrest rates. RESULTS (a) Pretrial detention averaged approximately 60 days compared with prevailing detentions averaging approximately 90 and 180 days in the same counties. (b) Pretrial rearrest rates were 17 percentage points higher than a similar pretrial population; however, representative comparison data are unavailable to confidently measure recidivism impacts. (c) Positive predictive power did not differ by race in predicting pretrial rearrests, SE = .04, 95% CI [.11, -.06], z = .61, p = .54, d = .08. (d) Despite comparable risk scores, African American participants were detained significantly longer than Caucasian participants (M = 60.92 vs. 45.58 days), p = .038, d = .18, 95% CI [.01, .36], and were less likely to receive a diversion opportunity (11% vs. 23%), p = .009, V = .17. CONCLUSION The observational design precludes causal conclusions; however, risk assessment was associated with shorter pretrial detention than prevailing bond practices with no racial disparities in risk prediction. Greater attention to risk assessment may reduce racial inequities in pretrial conditions. Representative comparison data are needed to measure the recidivism impacts of pretrial reform initiatives. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

中文翻译：

---

在审前释放决定中采用标准化的风险评估：与刑事司法结果和种族平等相关联。

目的我们研究了密西西比州法院为将审前释放决定基于风险评估而不是主要基于担保而做出的努力。假设（a）在同一县，审前拘留的时间将短于与现行债券惯例有关的拘留时间。（b）后座率将低于附近南部州的类似审前人口。（c）预测非裔美国人的假阳性的假阳性率将比白人参与者高。（d）由于较高的风险评分或评估覆盖率，非裔美国人参与者的审前拘留将更长。方法庭前被告（N = 521）在逮捕后的2周内完成了风险和需求分类（RANT），检查的结果包括庭前拘留的时间，索引病例的处置和再逮捕率。结果（a）在同一县，审前拘留平均约为60天，而普遍拘留平均约为90天和180天。（b）审前再逮捕率比类似的审前人口高17个百分点；但是，尚无代表性的比较数据来可靠地衡量累犯影响。（c）积极的预测能力在不同种族之间在预测审判前的后卫方面没有差异，SE = .04，95％CI [.11，-.06]，z = .61，p = .54，d = .08。（d）尽管风险评分相当，但非洲裔美国人的被拘留时间明显长于白人（M = 60.92比45.58天），p = .038，d = .18，95％CI [.01，.36]，接受转移的可能性较小（11％比23％），p = .009，V = .17。结论观察设计排除了因果关系的结论。但是，风险评估与审前羁押的时间短于现行的担保行为有关，并且在风险预测上没有种族差异。更加重视风险评估可以减少审前条件下的种族不平等。需要有代表性的比较数据来衡量审前改革措施的累犯性影响。（PsycInfo数据库记录（c）2020 APA，保留所有权利）。