NCJ Number
217315
Date Published
July 2006
Length
11 pages
Annotation
This bulletin presents the results of a validation test applied to the matching criteria used by New South Wales' (Australia) Reoffending Database (ROD) in linking the court records of individuals.
Abstract
The testing found that, despite noted limitation, the validation process found the ROD matching process to be highly reliable, which means that estimates on reoffending generated from ROD are likely to be sufficiently accurate for statistical and research purposes. The testing found that for every 10,000 people in the ROD, only 3 were incorrectly matched to another person. Although the estimated rate of false negatives (6.2 percent) was considerably higher than the estimated rate of false positives (0.057 percent), in most applications of ROD it is better to miss a match than to mistakenly make a match. ROD decides whether the defendants in two or more court records are likely to be the same individual by comparing their personal identifying information against five sets of matching criteria. If two court-appearance records match on at least one of these sets of criteria, they are considered to involve the same person. If not, they are viewed as distinct persons. To be matched under the ROD matching criteria, two records must have the same surname, first name, and data of birth; or surname, first name, middle name, and two components of the date of birth; or Central Names Index (CNI) number and date of birth; or CNI, surname, and two components of date of birth; or CNI, first name, and two components of the date of birth. The validation process first applied these matching criteria to a group of individuals to determine the frequency with which matching criteria generated false positives. It then created a "virtual" ROD to estimate the frequency of false negatives. 2 tables, 3 figures, and 16 notes