NCJ Number
204584
Date Published
November 1999
Length
103 pages
Annotation
This 1989 study of New York City's risk-assessment scheme for making pretrial-release decisions was designed to assess the predictive ability of the current point scale; to identify other predictors of failure-to-appear (FTA); and to formulate other alternative risk-classification schemes based on potentially new predictors.
Abstract
The first analytical phase of the study focused on FTA in criminal courts, and the second phase of the analysis addressed FTA among defendants under the supreme court jurisdiction. In the final phase of the study, FTA was examined without regard to the court of jurisdiction (combined-court analysis). Data were obtained from a random sample of 15,359 arrests made in 1989 in which defendants were held by the police until criminal court arraignment. The arrest-based sample was converted into a defendant-based sample (n=14,380), in which the defendant's first arrest during the sampling period was included in the study sample. The dependent variable, a measure of whether or not the defendant failed to appear, was defined as the issuance of a bench warrant at any appearance prior to the disposition of a defendant's case in the court under study. The FTA rate for defendants at risk under criminal court jurisdiction was 30.5 percent, and the supreme-court sample had a FTA rate of 33.1 percent. Without regard to the court of disposition, the FTA rate was 35.2 percent. A number of variables were examined for their statistical effects on pretrial FTA. These included community-ties items, criminal history indicators, severity and type of the top arrest charge, a defendant's socio-demographic attributes, and case-processing characteristics. Community-ties items included whether defendants had a working telephone in the residence; the length of time at the current address; whether the defendants had a New York City area address; family ties within the residence; whether someone was expected at the criminal court arraignment; and whether they were either employed, in school, or in a training program full time at the time of their arrest. the primary task in each analytical phase was to identify significant predictors of pretrial FTA. Logistic regression was used to analyze the data. A number of prediction models were constructed. With the exception of the supreme court analysis, the most predictive model from each analytical phase was used to guide the construction of a new point scale. Four risk-classification schemes were developed by dividing the sample of defendants at various points on the new scale. Scheme 4 was able to classify a larger proportion of defendants as good risks for a release-on-recognizance recommendation, while keeping the FTA at the same level. This was true for both the criminal court and the combined-court. The combined-court scheme had the added advantage of identifying at-risk defendants regardless of court of disposition. 21 tables, appended supplementary information, and a 20-item bibliography