December 2, 2009

III. Methodology

This report is based on 81 interviews conducted by Human Rights Watch with non-citizen detainees in Texas, Arizona, and New Mexico; detainees’ family members, immigrants’ rights advocates, and attorneys located throughout the United States; and ICE officials located in Washington, DC, Arizona, and Texas. Human Rights Watch also reviewed 158 pages of correspondence between ICE and detainees, their family members, and their congressional representatives, which were produced for Human Rights Watch by ICE in response to our Freedom of Information Act (FOIA) request.

The data on transfers (hereinafter “transfers dataset”) were obtained by Human Rights Watch from ICE on September 29, 2008, in response to a request we filed on February 27, 2008, under the Freedom of Information Act.[6] The numbers were analyzed by the Transactional Records Access Clearinghouse at Syracuse University.

The files released by ICE contain basic information concerning each exit of a detainee from a detention facility during the period October 1, 1998, through mid-April 2008. This information includes the nationality and gender of the detainee, the facility in which he or she had been detained, the ICE regional hub (known as a “docket control office” or “DCO”), and the dates of entry to and exit from this particular facility, as well as the date on which the immigrant had first been detained. A code also identified the exit reason such as “deported” or “removed,” “voluntary departure,” or “transfer.” However, no information concerning the reason for a transfer was provided.

The first step in TRAC’s research was to develop an analysis database. Initially, this required processing the 68 separate data files that had been released (each containing tens of thousands of records) and combining them into a single database of 3,376,269 records for further analysis. In addition, supplemental translation databases were prepared to map each coded entry to its definition. Consistency checks were also run against available published data. Finally, we checked each record for missing data and for undefined codes to minimize data entry errors.

TRAC also gathered additional information to classify each of the 1,524 detention facilities that appeared in the data. TRAC had previously obtained information on some of the facilities through separate research. For the remaining facilities, TRAC conducted telephone interviews and sought out other publicly available sources identifying the nature of each facility. Using this information each of the detention facilities was classified into broad categories, including “Service Processing Centers” (ICE owned and operated), “Intergovernmental Service Agreement facilities” (state and local jails under contract with ICE), “private contract detention facilities,” “federal Bureau of Prisons facilities” (under contract with ICE), “Office of Refugee Resettlement facilities” (under contract with ICE), and “Detention and Removal Operations juvenile facilities” (ICE owned and operated). Based upon address, each detention facility was also classified by state and by the federal court circuit in which it was located.

Additional analysis variables were then added to the database. For example, using the information on recorded dates, TRAC was able to compute the length of stay in a facility (number of days) and the fiscal year in which the transfer took place. Using this information along with the reasons a detainee was released from (“exited”) a detention facility, TRAC was able to classify records into those facilities where a detainee was placed on the initial day of detention (“originating” facilities) versus facilities to which the detainee was later transferred (“receiving” facilities).

The data also reveal that often a chain of transfers occurred. For example, records show that many immigrants were transferred to a facility and then shortly thereafter transferred out of the facility to another detention location. Unfortunately, it was not possible to match the transfers concerning the same individual because the files for the most part did not identify the particular detainee involved. As a result, while it was possible to classify in aggregate the originating and receiving detention facilities, it was not possible to directly connect the originating and receiving facility on individual transfers since each record only identified the originating detention facility and did not identify the particular facility to which a detainee was transferred.

Once the analysis database was developed, the actual analysis was carried out in two phases. The focus of the first phase was on the detainee population and transfer trends. All records where the exit reason was recorded as a transfer were included in this phase of the analysis. TRAC first examined changes over time in the volume of transfers. Second, TRAC analyzed national origin, gender, and other characteristics of the transferred detainee population and assessed whether there were any significant changes in the make-up of this population over time.

The second phase of the analysis focused upon the geographic location and other characteristics of the detention facilities, for both the originating facility and the receiving facility for the transfer. While it is known that transfers occur for many reasons, there was no information on why an individual transfer took place. For example, a transfer may occur to move a detainee close to the deportation location just prior to the detainee’s removal, or a detention facility may serve only as a convenient stopover between the originating and the intended destination facility. At some locations, ICE has specialized facilities that play a role in the intake process so that on initial pickup an immigrant may pass through more than one detention facility as part of the routine intake process.

While it would have been desirable to exclude these types of transfers from the analysis since they were not the focus of the study, there was no direct way to identify such records because the reason for the transfer was not given. However, it was possible to identify transfers involving “transient” stays—detention facilities in which the immigrant did not remain overnight. As a partial control, the set of receiving detention facilities analyzed in this phase of the research excluded any record where the immigrant arrived and left on the same day (“zero-day stays”) since these types of transfers clearly were outside the focus of this research. Similarly, the set of originating facilities excluded transfers within the same DCO that involved a zero-day stay to reduce double-counting of originating facilities where the intake process during the same day involved multiple facilities.

The resulting sets of originating and receiving detention facilities were then separately analyzed. For each set, facilities were ranked by the volume of transfers. Counts and rankings for originating and receiving detention facilities were also developed by type and by geographic location (state as well as federal court circuit).


[6] Letter from Catrina M. Pavlik-Keenan, FOIA officer, US Immigration and Customs Enforcement, US Department of Homeland Security, to Human Rights Watch, September 29, 2008 (letter on file with Human Rights Watch and reproduced in the Appendix to this report).