How did DHS estimate the size of the undocumented population during 2001–2008, and what were the official estimates?
Executive summary
The Department of Homeland Security (DHS) relied on a “residual” method between 2001 and 2008 that subtracted known legal foreign‑born counts from survey totals, adjusted for undercount, mortality and migration, to infer the unauthorized population (residual method) [1] [2]. Official DHS releases showed the undocumented population rising sharply in the early 2000s—DHS and related analyses put the count in the mid‑to‑low double‑digit millions, with commonly cited figures of roughly 8.5 million in 2000 rising to over 11 million by 2006 and Pew/DHS estimates near 12–12.4 million in 2007–2008 [3] [4] [5].
1. What “residual” means in practice: subtracting legal migrants from survey totals
DHS’s core approach began with a benchmark of the total foreign‑born population from Census Bureau sources—the 2000 decennial census early in the decade and the American Community Survey for later years—and then subtracted administrative counts of legal immigrants (naturalized citizens, lawful permanent residents, refugees, asylees and legal temporary residents) drawn from DHS and other federal records to produce a residual that is interpreted as the unauthorized population [2] [1] [6].
2. The crucial adjustments: undercount, mortality, emigration and status change
That raw residual is not the final number; DHS adjusted for undercount in household surveys (DHS commonly assumed about a 10 percent undercount for undocumented persons), for deaths using life‑table survival rates, and for emigration, removals and status adjustments (for example, unauthorized immigrants obtaining legal status) so that the annual series reflects flows in and out of the population as well as survey coverage errors [3] [2] [7].
3. Why DHS and others still publish similar numbers: methodology convergence
The residual approach developed by Warren and Passel and refined by DHS has become the dominant technique used by DHS, the Migration Policy Institute, the Congressional Budget Office and Pew; these independent analysts use similar inputs (Census/ACS counts and DHS administrative data) and similar adjustments, which explains why their point estimates are often close even as methodological nuances differ [2] [8].
4. Alternatives and probes of uncertainty: grouped answers, death‑registration and divergent methods
Government and academic reviewers have proposed or tested alternatives—GAO examined a “grouped answers” survey module to solicit immigration status indirectly, and researchers have used death‑registration and other models to probe coverage error and mortality differences—because residual estimates depend on assumptions about undercount and emigration that are hard to validate [9] [10] [7]. Some alternative modeling approaches produce much larger totals, underscoring that uncertainty is methodological as well as statistical [8].
5. The DHS official time path, 2001–2008: growth then slowing and signs of decline
Using the residual framework, DHS estimated a rapid increase in the unauthorized population in the early 2000s—consistent with academic summaries that report growth from about 8.5 million in 2000 to over 11 million by 2006—and DHS/OIS and Pew published peaks around 12–12.4 million for 2007 and roughly 11.9 million for 2008 in cited reports [3] [4] [5]. Analysts later emphasized that reduced arrivals after 2000, increased departures and enforcement contributed to slowing growth and the modest declines reported between 2007 and 2009 [5] [11].
6. What DHS itself warns and what remains unresolved
DHS documentation and academic reviews explicitly describe the component assumptions—coverage error, emigration rates, mortality, and administrative counts—and note sampling error and sensitivity to undercount assumptions; DHS and reviewers caution that year‑to‑year changes can reflect sampling and methodological choices as much as real population swings [1] [7]. The available sources document DHS’s methods and the headline estimates for 2001–2008 but do not resolve all disputes among alternative estimators, which is why independent corroboration and methodological transparency remain central to interpreting the official series [2] [8].