Immigration Court Case Outcomes in Georgia

Limitaciones

GEOGRAPHY IS THE ADDRESS OF RECORD, AND FOR DETAINED RESPONDENTS THAT IS THE DETENTION CENTER. `county_fips` is the county the Deportation Data Project derived from the ZIP code on the case; a detained respondent's address of record is the facility holding them. Stewart County's rows are 99.3% custody = 'detained_throughout' (Stewart Detention Center, Lumpkin), Charlton's 98.6% (the Folkston processing centers) and Irwin's 98.3% (Irwin County Detention Center) - these are not places those respondents live (Stewart County has under 6,000 residents). Use custody = 'never_detained' for a residence-like county map, and never read the custody = 'all' row of a facility county as residents; Fulton's 'all' row also mixes the Atlanta detention population with residents. `year` is the calendar year the case was FILED (the Notice to Appear date), not the year it was decided: recent filing years mix decided and pending cases (58% of 2026 filings and 26% of all Georgia filings in the window were still pending on 2026-08-01). `removal_order_rate` is over DECIDED cases only and will drift as pending cases resolve, and `avg_days_to_decision` is right-censored - only finished cases count, so the most recent years look fast; an apparent speed-up near the end of the window is a censoring artifact before it is a change in practice. Calendar years, not federal fiscal years: figures here will NOT match EOIR's own or DDP's published FY totals (federal FY runs October 1 - September 30). Coverage is the release window 1997-10-01 through 2026-08-01: calendar 1997 is October-December only and 2026 stops on 1 August, so both end years are partial and a year-over-year chart dips at both ends. 0 Georgia case(s) fell outside the window. The 29-year span is not a uniform series: filings swing from a 2020 trough (court closures) to a 2023 peak; EOIR's recording practice changed over the period and several fields are sparse in the early years; and the 2019-2026 surge is dominated by never-detained respondents whose cases are largely still open. Disclosure control (the platform's own rule; the source has none): county rows are annual (`month` is published at state grain only) with `demographic`, `in_absentia` and `citizenship_country` pinned to `all` and only `custody` broken out, on a complete 159-county x 30-year x 4-custody-value grid; any county count below 5 - ZERO INCLUDED - is withheld as NULL, with complementary suppression: across the custody partition the county `all` TOTAL is withheld whenever any custody part is withheld, so a published county total always equals the sum of its published parts and a withheld total reveals nothing about them (a facility county can show its never_detained residents while its misleading total is blank); across counties within a year; and of the removal count when the non-removal remainder would be 1-4. The still-pending remainder num_cases - num_cases_decided is NOT treated as a disclosive complement (a pending count is a property of the docket, not of a person; the same decision as the detention topics). A county NULL therefore means 'fewer than five, possibly none', never a real zero; state rows are complete and a state 0 is a real zero. avg_days_to_decision is additionally withheld, at EVERY grain, wherever fewer than 5 cases in the cell have been decided - an average over one to four people is those people's own durations. The 3 Georgia cases with no recorded custody status are counted in state rows only, so every county `all` total is exactly the sum of its custody parts. 143 of 4,770 county-year custody = 'all' cells and 1,650 custody = 'never_detained' cells publish num_cases. `demographic` is EOIR's recorded sex (agency-recorded, not self-reported), missing on 39.7% of Georgia cases, which count toward `all` only, so male + female is well below `all` - and the gap is an ERA effect, not random: sex is recorded on 3% of cases filed through 2012, 40% of 2013-2015 filings and 87% from 2016 on, so a male/female series is only meaningful from about 2016. 1,482 Georgia cases are 'Withholding Only' proceedings (case_type), which decide relief on an existing removal order and therefore can never produce a new removal order; they are counted in num_cases and num_cases_decided (their outcomes are 'Deny' / 'Grant' rather than 'Remove'), which lowers removal_order_rate very slightly. `citizenship_country` is EOIR's `nationality` field mapped onto the sub_topic's shared 22-country vocabulary (the gold column keeps the shared name so the ICE and EOIR topics agree); 5.2% of Georgia cases fall in `other`. Removal orders include the INA-withholding, CAT-withholding and CAT-deferral variants of `Remove`: each is a removal order whose execution to a particular country is barred, and the respondent is still ordered removed. Voluntary departure, termination, dismissal, relief granted and administrative closure are decided but not removal orders. Overall, 68.0% of decided Georgia cases in the window ended in a removal order. Judge, court, hearing-location and place fields are never published: EOIR names individual immigration judges, and outcome rates by named judge would be a statement about a person, not about Georgia. This topic is family 1 of the four measure families in the EOIR case database; representation (`eoir_representation`), bond decisions (`eoir_bond_decisions`) and relief applications (`eoir_asylum_decisions`) are separate topics built on the same source.