Appendix B — Common data elements
BLN crime codes
The four-digit crime codes in the data generally refer to the FBI’s National Criminal Information Center’s classification of criminal offenses. An example is “0904: Homicide-Willful Kill-Non-family-Weapon”. There are some made-up codes in the ICE data. They refer to more general or specific versions of existing codes. For example, ICE has two codes for illegal entry in the US that are more specific than the ones used by the FBI.
We have also assigned a 2-digit code (“09: Homicide”) matching those NCIC categories.
In addition, there are five BLN-created codes (bln_special_code) created to facilitate common stories:
Violent: FBI-defined violent crime codesDrugs: FBI-defined “Dangerous Drug” codesTraffic: Anything in the general code for traffic offensesImmigration: Codes that refer only to the illegal entry, re-entry or use of false documents to gain entry. We did not include the immigration offenses that refer to trafficking.No criminal conviction: Anything that did not have a conviction in the table’s corresponding criminality indicator.
(Violent and drug crime codes were assigned based on the Social Security Administration’s listing, which refers to FBI categorizations.)
Simplified categories
At the risk of losing some detail, BLN has simplified some of the categories used in the original data to make for more straightforward analysis.
Case status
We’re unsure when the case status is assigned. The original categories have their genesis in immigration law pre- and post-1996. We have recoded them into easier-to-understand categories:
| Original | Recode |
|---|---|
| 0-Withdrawal Permitted - I-275 Issued | Voluntary return |
| 3-Voluntary Departure Confirmed | Voluntary return |
| 5-Title 50 Expulsion | Alien Enemies Act expulsion |
| 6-Deported/Removed - Deportability | Removed |
| 7-Died | Died |
| 8-Excluded/Removed - Inadmissibility | Removed |
| 9-VR Witnessed | Voluntary return |
| A-Proceedings Terminated | Released |
| ACTIVE | Active case |
| B-Relief Granted | Released |
| E-Charging Document Canceled by ICE | Released |
| L-Legalization - Permanent Residence Granted | Released |
| Z-SAW - Permanent Residence Granted | Released |
| (Missing) | Active case |
City, County and State assignments
Detention locations
The original data had no state, city or county – just the name of the detention center and the code used by ICE. We used the Vera Institute’s Detention Trends dataset availble on its Github repo to assign the locations.
Detainer locations
The detainer table contained a name and city for each detainer facility. These were geocoded using Google’s Geocoding API. In some cases, the API returned a specific rooftop address for the location based on the name, but in most cases it contained only a general location. These coordinates were coded into counties using Census TIGER data, through R’s tigris package. Each county and state was checked and corrected where necessary.1
Other documentation
Deportation data project’s original documentation for ICE data. In particular, the full “codebook” is the source for a lot of our understanding.
There are some definitions that are still meaningful in the paper “Piecing together the US Immigration Detention Puzzle One Night at a Time, Donald Kerwin and Daniela Alulema (Center for Migration Studies), Journal on Migration and Human Security, 2012.
The only serious mistakes were in New Jersey, where Google coded them in New York City.↩︎