Coordination of Investigations by Department of Justice Violent Crime Task Forces

Evaluation and Inspections Report I-2007-004
May 2007
Office of the Inspector General


Appendix II
Methodology to Identify Duplicate Arrests

To analyze task forces’ efforts to coordinate investigations involving the same target, we asked each component to provide the name, date of birth, Social Security Number, and FBI Number (identifiers) of every individual who was arrested by one of the task forces during FYs 2003 through 2005. Table 7 shows the search criteria used by the components to generate these lists of individuals.

Table 7: Components’ Search Criteria

Component Database Searched Data Provided Number of Arrests Identified
ATF
N-Force
ATF first identified all investigations in N-Force that had been flagged with a Violent Crime Impact Team code. For those investigations, ATF provided identifiers for all defendants who were either (1) arrested before 10/01/2005 or (2) whose investigations were closed before 10/01/2005 without arrest.

1,678*

DEA
Priority Target Activity Resource Reporting System (PTARRS)
The DEA first identified all Mobile Enforcement Team deployments completed during FYs 2003 through 2005. For those investigations, the DEA provided identifiers for all defendants who were either (1) arrested or (2) whose investigations were closed before 10/01/2005 without arrest.

4,024

FBI
Integrated Statistical Reporting and Analysis Application (ISRAA)
The FBI first reviewed its list of case classifications and identified 71 classifications that were most likely to have been investigated by a Safe Streets Task Force.** Within those classifications, the FBI provided identifiers for all defendants who were arrested during FYs 2003 through 2005 in a field division that included one or more Safe Streets Task Forces.

24,265

USMS
Warrant Information Network (WIN)
The USMS first identified all districts that participate in a Regional Fugitive Task Force. The USMS then identified all individuals who were arrested by the USMS during FYs 2003 through 2005 and whose warrants had been issued in one of those districts.

67,261


* The VCIT program began in June 2004. Therefore, ATF provided only 16 months of arrest data, while the other components provided 3 years of arrest data.

** FBI case classifications identify the type of crime being investigated. The case classifications identified for this analysis were part of the Violent Incident Crimes Program, Violent Gangs Program, Major Theft Program, or Fugitive Program.

We entered the identifiers of every individual who was arrested by one of the task forces during FYs 2003 through 2005 into SPSS 14.0 and used the “Identify Duplicate Cases” feature to identify possible overlaps among the components. The feature allows the user to identify one or more fields that SPSS will compare to determine matches. The content of the fields must be identical for SPSS to declare a match. To ensure that differences in formatting did not lead SPSS to reject matching cases, we reformatted the lists from the components before importing them into SPSS.

We compared the lists to each other in the following pairs: USMS v. FBI, USMS v. DEA, USMS v. ATF, FBI v. DEA, FBI v. ATF, and DEA v. ATF. At least two “Identify Duplicate Cases” analysis runs were performed on each pair. First, each pair was compared using both the Name and Date of Birth fields. For cases to match on this analysis, both components needed to record the same name, using the same spelling, and record the same date of birth. We reviewed the results of each analysis and copied each pair of duplicate cases that contained one record from each component into a separate file for further analysis.32

The initial analysis could not identify matches involving names with different spellings (Steven v. Stephen), or matches where one component had recorded the defendant’s middle name while the other component had not. We performed a second “Identify Duplicate Cases” analysis run on each pair, comparing a different field, to identify cases that were not be captured in the first analysis.

An individual’s FBI Number is a unique identification number assigned to each individual who has a record in the FBI’s Integrated Automated Fingerprint Identification System (IAFIS), a nation-wide database of fingerprint and criminal history records of individuals who have been arrested. Every time an individual is arrested, he or she is fingerprinted, and the prints are submitted to IAFIS for comparison. If the prints match prints already in the system, the new arrest is appended to the existing criminal history record. We determined that a matching FBI Number represents the best possible way to identify a positive match between two records because it is tied to the fingerprint record and is therefore less susceptible to identity fraud than a date of birth or a Social Security Number.

For the second analysis run, we compared the components using the FBI Number field. The FBI Number comparison was performed on the USMS v. DEA, USMS v. ATF, and DEA v. ATF pairs. We could not perform this analysis on pairings that included the FBI because the FBI Number is not tracked in ISRAA and therefore was not included in the FBI’s data response. For the pairings including the FBI, we performed the second analysis run using the Social Security Number field. The Social Security Number comparison was done on the USMS v. FBI and FBI v. ATF pairs.33

We were concerned that our inability to compare FBI and DEA arrests on any field besides name and date of birth might lead us to underestimate the number of potential duplicates between those two components. At our request, employees of the FBI’s Criminal Justice Information Services Division (CJIS) conducted an offline search of the National Crime Information Center’s (NCIC) data for the FBI Numbers of the individuals identified in the FBI’s data response.34 We sent an exact copy of the FBI’s original ISRAA data to CJIS, which returned the data with an FBI Number field added to each record. Once we had received this information, we performed a final analysis on the USMS v. FBI, FBI v. DEA, and FBI v. ATF pairs, using the FBI Number for comparison.

We gave the components spreadsheets indicating which of their arrests were also reported by another component and asked them to provide details on the circumstances surrounding the arrests. We analyzed the components’ responses to determine the number of arrests reported by multiple components that were the results of joint investigations. Based on the components’ explanations, we developed 12 categories of reasons why arrests could be recorded by multiple components: 6 categories indicating cooperative efforts, 4 categories indicating duplication of effort without cooperation, and 2 neutral categories indicating other reasons for the duplication. We sorted the overlapping cases into these 12 categories and were able to determine the percentage of overlapping investigations that were the result of cooperation and the percentage of overlapping investigations resulting from duplication of effort.

We also asked the components to provide identifiers for individuals under investigation by the task forces during FYs 2003 through 2005 in which no arrest has been made. ATF, the DEA, and the USMS were able to provide this information, but the FBI could not. The FBI provided the identifiers of individuals arrested by FBI Safe Streets Task Forces from its Integrated Reporting and Statistical Software database, but this database does not contain information on investigations in which no arrest has been made. A random check of the identifiers for individuals under investigation provided by the FBI from the FBI Automated Casefile System, which does maintain information on individuals prior to arrest, showed the information provided by the FBI was not sufficient for comparison. Because FBI investigations accounted for 80 percent of the arrests reported by more than one task force we concluded that attempting to identify the suspects under investigation by more than one task force using identifiers from the FBI Automated Casefile System would not have any efficacy.



Footnotes
  1. Most of the matches identified by SPSS were internal duplicates, where a single component arrested the same individual more than once. Because our review did not focus on the potential for intra-component duplication, we did not include these matches in our analysis.

  2. The DEA’s list of arrestees did not include Social Security Numbers because that information was not tracked in the PTARRS database. As a result, only the Name and Date of Birth analysis was performed on the FBI v. DEA pair at this stage.

  3. NCIC is a database that provides information on criminal histories and open warrants for police officers and special agents. The FBI Number is one of the identifiers stored in this database. NCIC is commonly used by police officers during traffic stops, allowing them to learn if the driver is wanted or if the vehicle is stolen. It is also used by police officers and Special Agents at various stages of an investigation to determine whether a suspect has a criminal history.



« Previous Table of Contents Next »